Much of federated learning (FL) focuses on settings where local dataset statistics remain the same between training and testing. However, this assumption often does not hold in practice due to distribution shifts, motivating the development of domain generalization (DG) approaches that leverage source domain data to train models capable of generalizing to unseen target domains. In this paper, we are motivated by two major gaps in existing work on FL and DG: (1) the lack of formal mathematical analysis of DG objectives; and (2) DG research in FL being limited to the star-topology architecture. We develop Decentralized Federated Domain Generalization with Style Sharing ($\textit{StyleDDG}$), a decentralized DG algorithm which allows devices in a peer-to-peer network to achieve DG based on sharing style information inferred from their datasets. Additionally, we provide the first systematic approach to analyzing style-based DG training in decentralized networks. We cast existing centralized DG algorithms within our framework, and employ their formalisms to model $\textit{StyleDDG}$. We then obtain analytical conditions under which convergence of $\textit{StyleDDG}$ can be guaranteed. Through experiments on popular DG datasets, we demonstrate that $\textit{StyleDDG}$ can obtain significant improvements in accuracy across target domains with minimal communication overhead compared to baseline decentralized gradient methods.

It is well established that the performance of reconfigurable intelligent surface (RIS)-assisted systems critically depends on the optimal placement of the RIS. Previous works consider either simple coverage maximization or simultaneous optimization of the placement of the RIS along with the beamforming and reflection coefficients, most of which assume that the location of the RIS, base station (BS), and users are known. However, in practice, only the spatial variation of user density and obstacle configuration are likely to be known prior to deployment of the system. Thus, we formulate a non-convex problem that optimizes the position of the RIS over the expected minimum signal-to-interference-plus-noise ratio (SINR) of the system with user randomness, assuming that the system employs joint beamforming after deployment. To solve this problem, we propose a recursive coarse-to-fine methodology that constructs a set of candidate locations for RIS placement based on the obstacle configuration and evaluates them over multiple instantiations from the user distribution. The search is recursively refined within the optimal region identified in each stage to determine the final optimal region for RIS deployment. Detailed numerical results are presented to corroborate our findings.

Multimodal federated learning (MFL) aims to enrich model training in FL settings where clients are collecting measurements across multiple modalities. However, key challenges to MFL remain unaddressed, particularly in heterogeneous network settings where: (i) the set of modalities collected by each client is diverse, and (ii) communication limitations prevent clients from uploading all their locally trained modality encoders to the server. In this paper, we propose Multimodal Federated learning with joint Modality and Client selection (MFedMC), a communication-efficient MFL framework that tackles these challenges through a decoupled architecture and selective uploading. Unlike traditional holistic fusion approaches, MFedMC separates modality encoders and fusion modules: modality encoders are aggregated at the server for generalization across diverse client distributions, while fusion modules remain local to each client for personalized adaptation to individual modality configurations and data characteristics. Building on this decoupled design, our joint selection algorithm incorporates two main components: (a) A modality selection methodology for each client, which weighs (i) the impact of the modality, gauged by Shapley value analysis, (ii) the modality encoder size as a gauge of communication overhead, and (iii) the frequency of modality encoder updates, denoted recency, to enhance generalizability. (b) A client selection strategy for the server based on the local loss of modality encoders at each client. Experiments on five real-world datasets demonstrate that MFedMC achieves comparable accuracy to several baselines while reducing communication overhead by over 20×. A demo video and our code are available athttps://liangqiy.com/mfedmc/.

Federated learning (FL) encounters scalability challenges when implemented over fog networks that do not follow FL’s conventional star topology architecture. Semi-decentralized FL (SD-FL) has proposed a solution for device-to-device (D2D) enabled networks that divides model cooperation into two stages: at the lower stage, D2D communications is employed for local model aggregations within subnetworks (subnets), while the upper stage handles device-server (DS) communications for global model aggregations. However, existing SD-FL schemes are based on gradient diversity assumptions that become performance bottlenecks as data distributions become more heterogeneous. In this work, we develop semi-decentralized gradient tracking (SD-GT), the first SD-FL methodology that removes the need for such assumptions by incorporating tracking terms into device updates for each communication layer. Our analytical characterization of SD-GT reveals upper bounds on convergence for non-convex, convex, and strongly-convex problems. We show how the bounds enable the development of an optimization algorithm that navigates the performance-efficiency trade-off by tuning subnet sampling rate and D2D rounds for each global training interval. Our subsequent numerical evaluations demonstrate that SD-GT obtains substantial improvements in trained model quality and communication cost relative to baselines in SD-FL and gradient tracking on several datasets.

The traditional goal in remote tracking of a dynamic source is to keep the current estimate at the destination as close as possible to the true state. However, in domains such as surveillance applications, the destination is also interested in reconstructing the past trajectory of states for further processing. This requires striking a balance between providing current versus past state information so that the destination can optimize the trade-off between the metrics of freshness and reconstruction queue length. In this work, we propose a randomized update policy that decides between head-of-line versus tail-of-line packets in the update queue. As such, our policy combines the strength of Last-Come-First-Serve (LCFS) service discipline (which aims at reducing the age) with the strength of First-Come-First-Serve (FCFS) service discipline (which aims at reducing the reconstruction delay). We evaluate the performance of our proposed policy in terms of its randomization parameter, which can be optimized given the system parameters to achieve a better trade-off.

Remote tracking systems play a critical role in applications such as IoT, monitoring, surveillance and healthcare. In such systems, maintaining both real-time state awareness (for online decision making) and accurate reconstruction of historical trajectories (for offline post-processing) are essential. While the Age of Information (AoI) metric has been extensively studied as a measure of freshness, it does not capture the accuracy with which past trajectories can be reconstructed. In this work, we investigate reconstruction error as a complementary metric to AoI, addressing the trade-off between timely updates and historical accuracy. Specifically, we consider three policies, each prioritizing different aspects of information management: Keep-Old, Keep-Fresh, and our proposed Inter-arrival-Aware dropping policy. We compare these policies in terms of impact on both AoI and reconstruction error in a remote tracking system with a finite buffer. Through theoretical analysis and numerical simulations of queueing behavior, we demonstrate that while the Keep-Fresh policy minimizes AoI, it does not necessarily minimize reconstruction accuracy. In contrast, our proposed Inter-arrival-Aware dropping policy dynamically adjusts packet retention decisions based on generation times, achieving a balance between AoI and reconstruction error. Our results provide key insights into the design of efficient update policies for resource-constrained IoT networks.

Federated learning (FL) is vulnerable to backdoor attacks, where adversaries alter model behavior on target classification labels by embedding triggers into data samples. While these attacks have received considerable attention in horizontal FL, they are less understood for vertical FL (VFL), where devices hold different features of the samples, and only the server holds the labels. In this work, we propose a novel backdoor attack on VFL which (i) does not rely on gradient information from the server and (ii) considers potential collusion among multiple adversaries for sample selection and trigger embedding. Our label inference model augments variational autoencoders with metric learning, which adversaries can train locally. A consensus process over the adversary graph topology determines which datapoints to poison. We further propose methods for trigger splitting across the adversaries, with an intensity-based implantation scheme skewing the server towards the trigger. Our convergence analysis reveals the impact of backdoor perturbations on VFL indicated by a stationarity gap for the trained model, which we verify empirically as well. We conduct experiments comparing our attack with recent backdoor VFL approaches, finding that ours obtains significantly higher success rates for the same main task performance despite not using server information. Additionally, our results verify the impact of collusion on attack performance.

While federated learning (FL) eliminates the transmission of raw data over a network, it is still vulnerable to privacy breaches from the communicated model parameters. Differential privacy (DP) is often employed to address such issues. However, the impact of DP on FL in multi-tier networks – where hierarchical aggregations couple noise injection decisions at different tiers, and trust models are heterogeneous across subnetworks–is not well understood. To fill this gap, we develop Multi-Tier Federated Learning with Multi-Tier Differential Privacy (M2FDP), a DP-enhanced FL methodology for jointly optimizing privacy and performance over such networks. One of the key principles ofM2FDPis to adapt DP noise injection across the established edge/fog computing hierarchy (e.g., edge devices, intermediate nodes, and other tiers up to cloud servers) according to the trust models in different subnetworks. We conduct a comprehensive analysis of the convergence behavior ofM2FDPunder non-convex problem settings, revealing conditions on parameter tuning under which the training process converges sublinearly to a finite stationarity gap that depends on the network hierarchy, trust model, and target privacy level. We show how these relationships can be employed to develop an adaptive control algorithm forM2FDPthat tunes properties of local model training to minimize energy, latency, and the stationarity gap while meeting desired convergence and privacy criterion. Subsequent numerical evaluations demonstrate thatM2FDPobtains substantial improvements in these metrics over baselines for different privacy budgets and system configurations.

In multi-cell wireless networks, providing intelligent services via federated learning (FL) becomes more challenging due to multi-level distribution shifts across clients and regions, as well as additional communication delays among edge and cloud servers. To address these issues, we propose SplitOMC, a split learning framework that integrates overlapping-area clients and a multi-exit neural architecture to jointly handle (i) client-preferred, (ii) out-of-preference, and (iii) out-of-region tasks. By strategically leveraging clients in overlapping regions, SplitOMC accelerates training without excessive backhaul communication, while maintaining both personalization and generalization. We theoretically analyze the convergence behavior of the proposed algorithm, ensuring performance stability under heterogeneous data and communication conditions. Extensive experiments on MNIST, CIFAR-10/100, and a real-world Jetson Nano testbed demonstrate that SplitOMC consistently achieves faster training and inference with improved accuracy compared to state-of-the-art methods. In particular, the framework shows robustness in resource-constrained and unstable network environments, highlighting its practical value for next-generation wireless intelligent services.

Quantum federated learning (QFL) combines the robust data processing of quantum computing with the privacy-preserving features of federated learning (FL). However, in large-scale wireless networks, optimizing sum-rate is crucial for unlocking the true potential of QFL, facilitating effective model sharing and aggregation as devices compete for limited bandwidth amid dynamic channel conditions and fluctuating power resources. This paper studies a novel sum-rate maximization problem within a muti-channel QFL framework, specifically designed for non-orthogonal multiple access (NOMA)-based large-scale wireless networks. We develop a sum-rate maximization problem by jointly considering quantum device’s channel selection and transmit power. Our formulated problem is a non-convex, mixed-integer non-linear programming (MINLP) challenge that remains non-deterministic polynomial time (NP)-hard even with specified channel selection parameters. The complexity of the problem motivates us to create an effective iterative optimization approach that utilizes the sophisticated quantum approximate optimization algorithm (QAOA) to derive high-quality approximate solutions. Additionally, our study presents the first theoretical exploration of QFL convergence properties under full device participation, rigorously analyzing real-world scenarios with nonconvex loss functions, diverse data distributions, and the effects of quantum shot noise. Extensive simulation results indicate that our multi-channel NOMA-based QFL framework enhances model training and convergence behavior, surpassing conventional algorithms in terms of accuracy and loss. Moreover, our quantum-centric joint optimization approach achieves more than a 100% increase in sum-rate while ensuring rapid convergence, significantly outperforming the state-of-the-arts.

We investigate a real-time remote inference system where multiple correlated sources transmit observations over a communication channel to a receiver. The receiver utilizes these observations to infer multiple time-varying targets. Due to limited communication resources, the delivered observations may not be fresh. To quantify data freshness, we employ the Age of Information (AoI) metric. To minimize the inference error, we aim to design a signal-agnostic scheduling policy that leverages AoI without requiring knowledge of the actual target values or the source observations. This scheduling problem is a restless multi-armed bandit (RMAB) problem with a non-separable penalty function. Unlike traditional RMABs, the correlation among sources introduces a unique challenge: the penalty function of each source depends on the AoI of other correlated sources, preventing the problem from decomposing into multiple independent Markov Decision Processes (MDPs), a key step in applying traditional RMAB solutions. To address this, we propose a novel approach that approximates the penalty function for each source and establishes an analytical bound on the approximation error. We then develop scheduling policies for two scenarios: (i) full knowledge of the penalty functions and (ii) no knowledge of the penalty functions. For the case of known penalty functions, we present an upper bound on the optimality gap that highlights the impact of the correlation parameter and the system size. For the case of unknown penalty functions and signal distributions, we develop an online learning approach that utilizes bandit feedback to learn an online Maximum Gain First policy. Simulation results demonstrate the effectiveness of our proposed policies in minimizing inference error and achieving scalability in the number of sources.

Large Language Models (LLMs) can perform zero-shot learning on unseen tasks and few-shot learning on complex reasoning tasks. However, resource-limited mobile edge networks struggle to support long-context LLM serving for LLM agents during multi-round interactions with users. Unlike stateless computation offloading and static service offloading in edge computing, optimizing LLM serving at edge servers is challenging because LLMs continuously learn from context which raises accuracy, latency, and resource consumption dynamics. In this paper, we propose a joint model caching and inference offloading framework that utilizes test-time deep reinforcement learning (T2DRL) to optimize deployment and execution strategies for long-context LLM serving. In this framework, we analyze the performance convergence and design an optimization problem considering the utilization of context windows in LLMs. Furthermore, the T2DRL algorithm can learn in both the training phase and the testing phase to proactively manage cached models and service requests and adapt to context changes and usage patterns during execution. To further enhance resource allocation efficiency, we propose a double Dutch auction (DDA) mechanism, which dynamically aligns the marginal value of an additional reasoning path with the marginal cost of reasoning services. Finally, experimental results demonstrate that the T2DRL algorithm can reduce system costs by at least 30% compared to baselines while guaranteeing the performance of LLM agents in real-world perception and reasoning tasks.

Federated learning (FL) is a popular technique for distributing machine learning (ML) across a set of edge devices. In this paper, we study fully decentralized FL, where in addition to devices conducting training locally, they carry out model aggregations via cooperative consensus formation over device-to-device (D2D) networks. We introduce asynchronous, event-triggered communications among the devices to handle settings where access to a central server is not feasible. To account for the inherent resource heterogeneity and statistical diversity challenges in FL, we define personalized communication triggering conditions at each device that weigh the change in local model parameters against the available local network resources. We theoretically recover the$\mathcal {O}{(\ln {k} / \sqrt {k})}$convergence rate to the globally optimal model of decentralized gradient descent (DGD) methods in the setup of our methodology. We provide our convergence guarantees for the last iterates of models, under relaxed graph connectivity and data heterogeneity assumptions compared with the existing literature. To do so, we demonstrate a$B$-connected information flow guarantee in the presence of sporadic communications over the time-varying D2D graph. Our subsequent numerical evaluations demonstrate that our methodology obtains substantial improvements in convergence speed and/or communication savings compared to existing decentralized FL baselines.

Interest continues to grow in utilizing federated learning (FL) for various signal processing and communications applications. Over-the-air (OTA) computation has been proposed to improve FL efficiency in bandwidth-limited environments by leveraging the superposition characteristic of a wireless multiple-access channel (MAC). However, OTA FL faces inherent challenges due to channel noise and fading in any wireless MAC scenario, which can degrade optimization and significantly reduce model accuracy. This paper aims to design a robust OTA FL system to counteract the effects of noise and fading over time-varying channels. We propose a novel approach employing a Kalman filter (KF)-based OTA FL algorithm under imperfect channel state information (CSI). We conduct a convergence analysis of our OTA FL scheme, which motivates our development of a complementary hierarchical optimization methodology to minimize the impact of bias and noise terms. Numerical results confirm that our methodology has superior performance to conventional OTA FL, and approaches the performance obtained by the upper limit of perfect CSI in low-SNR scenarios.

A few recent studies have shown the benefits of using centrally pre-trained models to initialize federated learning (FL). However, existing methods do not generalize well when faced with an arbitrary set of downstream FL tasks. Specifically, they often (i) achieve limited accuracy, especially with unseen downstream labels, and (ii) result in significant accuracy variance, failing to provide a balanced performance across clients. To address these challenges, we propose CoPreFL, a collaborative/distributed pre-training approach that robustly initializes for downstream FL tasks. CoPreFL leverages model-agnostic meta-learning (MAML) that tailors the global model to mimic heterogeneous and unseen FL scenarios, resulting in a pre-trained model that is rapidly adaptable to any FL task. Our MAML procedure integrates performance variance into the meta-objective function, balancing performance across clients rather than solely optimizing for accuracy. Extensive experiments show that CoPreFL significantly enhances average accuracy and reduces variance in arbitrary downstream FL tasks with unseen/seen labels, outperforming various pre-training baselines. Additionally, CoPreFL proves compatible with different well-known FL algorithms used in downstream tasks, boosting performance in each case.

The introduction of massive MIMO (Multiple Input Multiple Output) communication systems enables base stations (BS) to perform beamforming for enhancing communication reliability. A typical key assumption, however, is the availability of accurate downlink channel state information (CSI). In practice, CSI estimation and reporting delays coupled with the process of channel aging result in the BS receiving outdated CSI information, which in turn impacts the system's spectral efficiency. To combat this latency, this paper develops efficient methods of CSI prediction that preemptively predict future downlink CSI based on historical data. We leverage the spatial and temporal correlation properties of the channel and use explicit feature extraction frameworks for both dimensions to accurately predict future CSI. We analyze combinations of spatial and temporal feature extractors in terms of a tradeoff between performance and latency. We evaluate the performance of the proposed prediction model in terms of proximity to the ground truth, prediction latency, and model footprint. Our experiments show that our method outperforms classical statistical methods as well as existing CSI prediction baselines.

This paper investigates an OFDM-based over-the-air federated learning (OTA-FL) system, where multiple mobile devices, e.g., unmanned aerial vehicles (UAVs), transmit local machine learning (ML) models to a central parameter server (PS) for global model aggregation. The high mobility of local devices results in imperfect channel estimation, leading to a misalignment problem, i.e., the model parameters transmitted from different local devices do not arrive at the central PS simultaneously. Moreover, the mobility introduces time-varying uploading channels, which further complicates the aggregation process. All these factors collectively cause distortions in the OTA-FL training process which are underexplored. To quantify these effects, we first derive a closed-form expression for a single-round global model update in terms of these channel imperfections. We then extend our analysis to capture multiple rounds of global updates, yielding a bound on the accumulated error in OTA-FL. We validate our theoretical results via extensive numerical simulations, which corroborate our derived analysis.

We consider a setting where multiple correlated sources send real-time observations over a wireless communication channel to a receiver. The receiver uses the delivered observations to infer multiple time-varying targets. Due to limited communication resources, these observations may not always be fresh. To quantify data timeliness, we utilize the Age of Information (AoI) metric. Our goal is to minimize realtime inference error by developing signal-agnostic scheduling policies that leverage AoI without requiring knowledge of the actual target values or the specific source observations. For the two-source case, we obtain an optimal cyclic policy with low computational complexity. For more than two-sources, we establish an information-theoretic lower bound on inference error. Building upon this lower bound, we approximate the scheduling problem and propose an approximate Whittle index policy that is asymptotically optimal as the number of sources increases and the correlation among sources decreases. Our scheduling policies hold for arbitrary target and source processes and loss functions. Finally, we conduct simulations of a network of cameras with overlapping field of views tracking multiple mobile objects to demonstrate the effectiveness of our policies.

Over the past several years, various federated learning (FL) methodologies have been developed to improve model accuracy, a primary performance metric in machine learning. However, to utilize FL in practical decision-making scenarios, beyond considering accuracy, the trained model must also have a reliable confidence in each of its predictions, an aspect that has been largely overlooked in existing FL research. Motivated by this gap, we propose Non-Uniform Calibration for Federated Learning (NUCFL), a generic framework that integrates FL with the concept of model calibration. The inherent data heterogeneity in FL environments makes model calibration particularly difficult, as it must ensure reliability across diverse data distributions and client conditions. Our NUCFL addresses this challenge by dynamically adjusting the model calibration objectives based on statistical relationships between each client's local model and the global model in FL. In particular, NUCFL assesses the similarity between local and global model relationships, and controls the penalty term for the calibration loss during client-side local training. By doing so, NUCFL effectively aligns calibration needs for the global model in heterogeneous FL settings while not sacrificing accuracy. Extensive experiments show that NUCFL offers flexibility and effectiveness across various FL algorithms, enhancing accuracy as well as model calibration.

To improve the efficiency of reinforcement learning (RL), we propose a novel asynchronous federated reinforcement learning (FedRL) framework termed AFedPG, which constructs a global model through collaboration among $N$ agents using policy gradient (PG) updates. To address the challenge of lagged policies in asynchronous settings, we design a delay-adaptive lookahead technique *specifically for FedRL* that can effectively handle heterogeneous arrival times of policy gradients. We analyze the theoretical global convergence bound of AFedPG, and characterize the advantage of the proposed algorithm in terms of both the sample complexity and time complexity. Specifically, our AFedPG method achieves $\mathcal{O}(\frac{{\epsilon}^{-2.5}}{N})$ sample complexity for global convergence at each agent on average. Compared to the single agent setting with $\mathcal{O}(\epsilon^{-2.5})$ sample complexity, it enjoys a linear speedup with respect to the number of agents. Moreover, compared to synchronous FedPG, AFedPG improves the time complexity from $\mathcal{O}(\frac{t_{\max}}{N})$ to $\mathcal{O}({\sum_{i=1}^{N} \frac{1}{t_{i}}})^{-1}$, where $t_{i}$ denotes the time consumption in each iteration at agent $i$, and $t_{\max}$ is the largest one. The latter complexity $\mathcal{O}({\sum_{i=1}^{N} \frac{1}{t_{i}}})^{-1}$ is always smaller than the former one, and this improvement becomes significant in large-scale federated settings with heterogeneous computing powers ($t_{\max}\gg t_{\min}$). Finally, we empirically verify the improved performance of AFedPG in four widely used MuJoCo environments with varying numbers of agents. We also demonstrate the advantages of AFedPG in various computing heterogeneity scenarios.

Decentralized federated learning (DFL) captures FL settings where both (i) model updates and (ii) model aggregations are exclusively carried out by the clients without a central server. Existing DFL works have mostly focused on settings where clients conduct a fixed number of local updates between local model exchanges, overlooking heterogeneity and dynamics in communication and computation capabilities. In this work, we propose Decentralized Sporadic Federated Learning ($\texttt{DSpodFL}$), a DFL methodology built on a generalized notion of *sporadicity* in both local gradient and aggregation processes. $\texttt{DSpodFL}$ subsumes many existing decentralized optimization methods under a unified algorithmic framework by modeling the per-iteration (i) occurrence of gradient descent at each client and (ii) exchange of models between client pairs as arbitrary indicator random variables, thus capturing *heterogeneous and time-varying* computation/communication scenarios. We analytically characterize the convergence behavior of $\texttt{DSpodFL}$ for both convex and non-convex models and for both constant and diminishing learning rates, under mild assumptions on the communication graph connectivity, data heterogeneity across clients, and gradient noises. We show how our bounds recover existing results from decentralized gradient descent as special cases. Experiments demonstrate that $\texttt{DSpodFL}$ consistently achieves improved training speeds compared with baselines under various system settings.

In recent years, machine learning (ML) methods have become increasingly popular in wireless communication systems for several applications. A critical bottleneck for designing ML systems for wireless communications is the availability of realistic wireless channel datasets, which are extremely resource-intensive to produce. To this end, the generation of realistic wireless channels plays a key role in the subsequent design of effective ML algorithms for wireless communication systems. Generative models have been proposed to synthesize channel matrices, but outputs produced by such methods may not correspond to geometrically viable channels and do not provide any insight into the scenario being generated. In this work, we aim to address both these issues by integrating established parametric, physics-based geometric channel (PPGC) modeling frameworks with generative methods to produce realistic channel matrices with interpretable representations in the parameter domain. We show that the generative model converges to prohibitively suboptimal stationary points when learning the underlying prior directly over the parameters due to the non-convex PPGC model. To address this limitation, we propose a linearized reformulation of the problem to ensure smooth gradient flow during generative model training, while also providing insights into the underlying physical environment. We evaluate our model against prior baselines by comparing the generated, scenario-specific samples in terms of the 2-Wasserstein distance and through its utility when used for downstream compression tasks.

Many intelligence tasks operate over networks where observations are vertically split across devices, necessitating collaborative inference. Existing collaborative learning approaches, such as Vertical Federated Learning (VFL), typically implicitly assume the existence of an architecture that is reasonably fault-tolerant, e.g., a star topology to an aggregator node that never fails. However, in practice, device networks may be decentralized and possess dynamic connectivity, making them susceptible to catastrophic faults (e.g., environmental disruptions, extreme weather). In this work, we study the problem of enabling robust collaborative inference over these decentralized, dynamic, and fault-prone networks. We first formulate the impact of faults on collaborative inference through a notion of dynamic risk for the data and network context. Then, we develop Multiple Aggregation with Gossip Rounds and Simulated Faults (MAGS) which synthesizes three features to enhance fault tolerance during inference: (i) fault simulation via dropout in training, (ii) replication of aggregators across devices, and (iii) gossip layers to produce an ensemble inference. We provide theoretical insights into why each of these components enhances robustness, e.g., proving that the gossip protocol reduces dynamic risk according to prediction diversity. We conduct extensive evaluations over five datasets and different network configurations, which validate that MAGS substantially improves robustness over VFL baselines. The code is available at: https://github.com/inouye-lab/MAGS_Distributed_Robust_Learning

Compared to traditional machine learning models, recent large language models (LLMs) can exhibit multi-task-solving capabilities through multiple dialogues and multi-modal data sources. These unique characteristics of LLMs, together with their large model size, make their deployment more challenging. Specifically, (i) deploying LLMs on local devices faces computational, memory, and energy resource issues, while (ii) deploying them in the cloud cannot guarantee real-time service and incurs communication/usage costs. In this paper, we design TMO, a local-cloud LLM inference system with Three-M Offloading: Multi-modal, Multi-task, and Multi-dialogue. TMO incorporates (i) a lightweight local LLM that can process simple tasks at high speed and (ii) a large-scale cloud LLM that can handle multi-modal data sources. We develop a resource-constrained reinforcement learning (RCRL) strategy for TMO that optimizes the inference location (i.e., local vs. cloud) and multi-modal data sources to use for each task/dialogue, aiming to maximize the long-term reward (response quality, latency, and usage cost) while adhering to resource constraints. We also contribute M4A1, a new dataset we curated that contains reward and cost metrics across multiple modality, task, dialogue, and LLM configurations, enabling evaluation of offloading decisions. We demonstrate the effectiveness of TMO compared to several exploration-decision and LLM-as-Agent baselines, showing significant improvements in latency, cost, and response quality.

As the era of autonomous agents making decisions on behalf of users unfolds, ensuring contextual integrity (CI) -- what is the appropriate information to share while carrying out a certain task -- becomes a central question to the field. We posit that CI demands a form of reasoning where the agent needs to reason about the context in which it is operating. To test this, we first prompt LLMs to reason explicitly about CI when deciding what information to disclose. We then extend this approach by developing a reinforcement learning (RL) framework that further instills in models the reasoning necessary to achieve CI. Using a synthetic, automatically created, dataset of only $\sim700$ examples but with diverse contexts and information disclosure norms, we show that our method substantially reduces inappropriate information disclosure while maintaining task performance across multiple model sizes and families. Importantly, improvements transfer from this synthetic dataset to established CI benchmarks such as PrivacyLens that has human annotations and evaluates privacy leakage of AI assistants in actions and tool calls.

While federated learning (FL) and differential privacy (DP) have been extensively studied, their application to automatic speech recognition (ASR) remains largely unexplored due to the challenges in training large transformer models. Specifically, large models further exacerbate issues in FL as they are particularly susceptible to gradient heterogeneity across layers, unlike the relatively uniform gradient behavior observed in shallow models. As a result, prior works struggle to converge with standard optimization techniques, even in the absence of DP mechanisms. To the best of our knowledge, no existing work establishes a competitive, practical recipe for FL with DP in the context of ASR. To address this gap, we establish **the first benchmark for FL with DP** in end-to-end ASR. Our approach centers on per-layer clipping and layer-wise gradient normalization: theoretical analysis reveals that these techniques together mitigate clipping bias and gradient heterogeneity across layers in deeper models. Consistent with these theoretical insights, our empirical results show that FL with DP is viable under strong privacy guarantees, provided a population of at least several million users. Specifically, we achieve user-level ($7.2$, $10^{-9}$)-DP (resp. ($4.5$, $10^{-9}$)-DP) with only a 1.3\% (resp. 4.6\%) absolute drop in word error rate when extrapolating to high (resp. low) population scales for FL with DP in ASR. Although our experiments focus on ASR, the underlying principles we uncover — particularly those concerning gradient heterogeneity and layer-wise gradient normalization — offer broader guidance for designing scalable, privacy-preserving FL algorithms for large models across domains. Code of all experiments and benchmarks is available at https://github.com/apple/ml-pfl4asr.

Machine learning (ML) algorithms deployed in real-world environments are often faced with the challenge of adapting models to concept drift, where the task data distributions are shifting over time. The problem becomes even more difficult when model performance must be maintained under adherence to strict resource constraints. Existing solutions often depend on drift-detection methods that produce high computational overhead for resource-constrained environments, and fail to provide strict guarantees on resource usage or theoretical performance assurances. To address these shortcomings, we propose RCCDA: a dynamic model update policy that optimizes ML training dynamics while ensuring compliance to predefined resource constraints, utilizing only past loss information and a tunable drift threshold. In developing our policy, we analytically characterize the evolution of model loss under concept drift with arbitrary training update decisions. Integrating these results into a Lyapunov drift-plus-penalty framework produces a lightweight greedy-optimal policy that provably limits update frequency and cost. Experimental results on four domain generalization datasets demonstrate that our policy outperforms baseline methods in inference accuracy while adhering to strict resource constraints under several schedules of concept drift, making our solution uniquely suited for real-time ML deployments.

Cooperative localization leverages noisy inter-node distance measurements and exchanged wireless messages to estimate node positions in a wireless network. In communicationconstrained environments, however, transmitting large messages becomes problematic. In this paper, we propose an approach for communication-efficient cooperative localization that addresses two main challenges. First, cooperative localization often needs to be performed over wireless networks with loopy graph topologies. Second is the need for designing an algorithm that has low localization error while simultaneously requiring a much lower communication overhead. Existing methods fall short of addressing these two challenges concurrently. To achieve this, we propose a vector quantized message passing neural network (VQ-MPNN) for cooperative localization. Through end-to-end neural network training, VQ-MPNN enables the co-design of node localization and message compression. Specifically, VQMPNN treats prior node positions and distance measurements as node and edge features, respectively, which are encoded as node and edge states using a graph neural network. To find an efficient representation for the node state, we construct a vector quantized codebook for all node states such that instead of sending long messages, each node only needs to transmit a codeword index. Numerical evaluations demonstrates that our proposed VQ-MPNN approach can deliver localization errors that are similar to existing approaches while reducing the overall communication overhead by an order of magnitude.

As federated learning (FL) becomes more widespread, there is growing interest in its decentralized variants. Decentralized FL leverages the benefits of fast and energy-efficient device-to-device communications to obviate the need for a central server. However, this opens the door to new security vulnerabilities as well. While FL security has been a popular research topic, the role of adversarial node placement in decentralized FL remains largely unexplored. This article addresses this gap by evaluating the impact of various coordinated adversarial node placement strategies on decentralized FL’s model training performance. We adapt two threads of placement strategies to this context: 1) maximum span-based algorithms and 2) network centrality-based approaches. Building on them, we propose a novel attack strategy, MaxSpAN-FL, which is a hybrid between these paradigms that adjusts node placement probabilistically based on network topology characteristics. Numerical experiments demonstrate that our attack consistently induces the largest degradation in decentralized FL models compared with baseline schemes across various network configurations and numbers of coordinating adversaries. We also provide theoretical support for why eigenvector centrality-based attacks are suboptimal in decentralized FL. Overall, our findings provide valuable insights into the vulnerabilities of decentralized FL systems, setting the stage for future research aimed at developing more secure and robust decentralized FL frameworks.

Although user cooperation cannot improve the capacity of Gaussian two-way channels (GTWCs) with independent noises, it can improve communication reliability. In this work, we aim to enhance and balance the communication reliability in GTWCs by minimizing the sum of error probabilities via joint design of encoders and decoders at the users. We first formulate general encoding/decoding functions, where the user cooperation is captured by the coupling of user encoding processes. The coupling effect renders the encoder/decoder design non-trivial, requiring effective decoding to capture this effect, as well as efficient power management at the encoders within power constraints. To address these challenges, we propose two different twoway coding strategies: linear coding and learning-based coding. For linear coding, we propose optimal linear decoding and discuss new insights on encoding regarding user cooperation to balance reliability. We then propose an efficient algorithm for joint encoder/decoder design. For learning-based coding, we introduce a novel recurrent neural network (RNN)-based coding architecture, where we propose interactive RNNs and a power control layer for encoding, and we incorporate bi-directional RNNs with an attention mechanism for decoding. Through simulations, we show that our two-way coding methodologies outperform conventional channel coding schemes (that do not utilize user cooperation) significantly in sum-error performance. We also demonstrate that our linear coding excels at high signal-to-noise ratios (SNRs), while our RNN-based coding performs best at low SNRs. We further investigate our two-way coding strategies in terms of power distribution, two-way coding benefit, different coding rates, and block-length gain.

The adversarial wiretap channel of type II (AWTC-II) is a communication channel that can a) read a fraction of the transmitted symbols up to a given bound and b) induce both errors and erasures in a fraction of the symbols up to given bounds. The channel is controlled by an adversary who can freely choose the locations of the symbol reads, errors and erasures via a process with unbounded computational power. The AWTC-II is an extension of Ozarow’s and Wyner’s wiretap channel of type II to the adversarial channel setting. The semantic-secrecy (SS) capacity of the AWTC-II is partially known, where the best-known lower bound is non-constructive and proven via a random coding argument that uses a large number (that is, exponential in blocklengthn) of random bits to describe the random code. In this work, we establish a new derandomization result in which we match the best-known lower bound via a non-constructive random code that uses onlyO(n2) random bits. Unlike fully random codes, our derandomized code admits an efficient encoding algorithm and benefits from some linear structure. Our derandomization result is a novel application ofrandom pseudolinear codes– a class of non-linear codes first proposed for applications outside the AWTC-II setting, which havek-wise independent codewords wherekis a design parameter. As the key technical tool in our analysis, we provide a novel concentration inequality for sums of random variables with limited independence, as well as a soft-covering lemma similar to that of Goldfeld, Cuff and Permuter that holds for random codes withk-wise independent codewords.

This paper delves into the design of an energy efficient collaborative federated learning (CFL) methodology using which mobile devices exchange their FL model with a subset of their neighbors without reliance on a parameter server based on the distributed graph neural network (GNN) method. Each device is unable to send its FL model to every neighboring device due to device mobility and wireless resource limitations. To reduce the energy consumption of FL model transmission, each device must choose a subset of devices with which to share its FL model. This problem is formulated as an optimization problem to meet the constraints of delay and training loss while minimizing the energy consumption for model transmission. However, the formulated problem is difficult to solve since the device mobility patterns, and the relationship between the device connection scheme and CFL performance are unknown. To address this challenge, we analytically characterize the relationship between dynamic device connections and the performance of CFL methodology. Based on the analysis, a GNN based algorithm is proposed to enable each device to select a subset of its neighbors and the transmit power in a decentralized method. Compared to standard optimization methods that must determine device connections in a centralized manner, the GNN based method enables each device to use its neighboring devices' location and connection information to individually determine a subset of devices to transmit the local model. Given the device connections, the optimal transmit power of each device can be determined by convex optimization. Simulation results show that the proposed method can reduce the energy consumption for model transmission and training loss by up to 46% and 2%, respectively

This article proposes a novel communication-efficient split learning (SL) framework, named SplitFC, which reduces the communication overhead required for transmitting intermediate features and gradient vectors during the SL training process. The key idea of SplitFC is to leverage different dispersion degrees exhibited in the columns of the matrices. SplitFC incorporates two compression strategies: 1) adaptive feature-wise dropout and 2) adaptive feature-wise quantization. In the first strategy, the intermediate feature vectors are dropped with adaptive dropout probabilities determined based on the standard deviation of these vectors. Then, by the chain rule, the intermediate gradient vectors associated with the dropped feature vectors are also dropped. In the second strategy, the non-dropped intermediate feature and gradient vectors are quantized using adaptive quantization levels determined based on the ranges of the vectors. To minimize the quantization error, the optimal quantization levels of this strategy are derived in a closed-form expression. Simulation results on the MNIST, CIFAR-100, and CelebA datasets demonstrate that SplitFC outperforms state-of-the-art SL frameworks by significantly reducing communication overheads while maintaining high accuracy.

Hierarchical federated learning (HFL) has demonstrated promising scalability advantages over the traditional “star-topology” architecture-based federated learning (FL). However, HFL still imposes significant computation, communication, and storage burdens on the edge, especially when training a large-scale model over resource-constrained wireless devices. In this paper, we propose hierarchical independent submodel training (HIST), a new FL methodology that aims to address these issues in hierarchical cloud-edge-client networks. The key idea behindHISTis to divide the global model into disjoint partitions (or submodels) per round so that each group of clients (i.e., cells) is responsible for training only one partition of the model. We characterize the convergence behavior ofHISTunder mild assumptions, showing the impacts of several key attributes (e.g., submodel sizes, number of cells, edge and global aggregation frequencies) on the rate and stationarity gap. Building upon the theoretical results, we propose a submodel partitioning strategy to minimize the training latency depending on network resource availability and a target learning performance guarantee. We then demonstrate howHISTcan be augmented with over-the-air computation (AirComp) to further enhance the efficiency of the model aggregation over the edge cells. Through numerical evaluations, we verify thatHISTis able to save training time and communication costs by wide margins while achieving comparable accuracy as conventional HFL. Moreover, our experiments demonstrate that AirComp-assistedHISTprovides further improvements in training latency.

Hierarchical federated learning (HFL) has demon-strated promising scalability advantages over the traditional “star-topology” architecture-based federated learning (FL). How-ever, HFL still imposes significant computation, communication, and storage burdens on the edge, especially when training a large-scale model over resource-constrained Internet of Things (IoT) devices. In this paper, we propose hierarchical independent submodel training (HIST), a new FL methodology that aims to address these issues in hierarchical settings. The key idea behind HIST is a hierarchical version of model partitioning, where we partition the global model into disjoint submodels in each round, and distribute them across different cells, so that each cell is responsible for training only one partition of the full model. This enables each client to save computation/storage costs while alleviating the communication loads throughout the hierarchy. We characterize the convergence behavior of HIST for non-convex loss functions under mild assumptions, showing the impact of several attributes (e.g., number of cells, local and global aggregation frequency) on the performance-efficiency tradeoff. Finally, through numerical experiments, we verify that HIST is able to save communication costs by a wide margin while achieving the same target testing accuracy.

While network coverage maps continue to expand, many devices located in remote areas remain unconnected to terrestrial communication infrastructures, preventing them from getting access to the associated data-driven services. In this paper, we propose a cooperative ground-to-satellite federated learning (FL) methodology to facilitate machine learning service management over remote regions. Our methodology orchestrates satellite constellations to provide the following key functions during FL: (i) processing data offloaded from ground devices, (ii) aggregating models within device clusters, and (iii) relaying models/data to other satellites via inter-satellite links (ISLs). Due to the limited coverage time of each satellite over a particular remote area, we facilitate satellite transmission of trained models and acquired data to neighboring satellites via ISL, so that the incoming satellite can continue FL for the region. We also develop a training latency minimizer which optimizes over the amount of data to be offloaded from ground devices to satellites. Through experiments on benchmark datasets, we show that our scheme can significantly speed up the convergence of FL compared with terrestrial-only and other satellite baseline approaches.

Dual-function radar-communication (DFRC) is a promising technology where radar and communication functions operate on the same spectrum and hardware. In this paper, we propose an algorithm for designing constant modulus waveforms for DFRC systems. Particularly, we jointly optimize the correlation properties and the spatial beam pattern. For communication, we employ constructive interference-based block-level precoding (CI-BLP) to exploit distortion due to multi-user and radar transmission. We propose a majorization-minimization (MM)-based solution to the formulated problem. To accelerate convergence, we propose an improved majorizing function that leverages a novel diagonal matrix structure. We then evaluate the proposed algorithm via comprehensive simulations.

One of the main challenges of decentralized machine learning paradigms such as Federated Learning (FL) is the presence of local non-i.i.d. datasets. Device-to-device transfers (D2D) between distributed devices has been shown to be an effective tool for dealing with this problem and robust to stragglers. In an unsupervised case, however, it is not obvious how data exchanges should take place due to the absence of labels. In this paper, we propose an approach to create an optimal graph for data transfer using Reinforcement Learning. The goal is to form links that will provide the most benefit considering the environment's constraints and improve convergence speed in an unsupervised FL environment. Numerical analysis shows the advantages in terms of convergence speed and straggler resilience of the proposed method to different available FL schemes and benchmark datasets.

A recent line of research has been investigating deep learning approaches to wireless positioning (WP). Although these WP algorithms have demonstrated high accuracy and robust performance against diverse channel conditions, they also have a major drawback: they require processing high-dimensional features, which can be prohibitive for mobile applications. In this work, we design a positioning neural network (P-NN) that substantially reduces the complexity of deep learning-based WP through carefully crafted minimum description features. Our feature selection is based on maximum power measurements and their temporal locations to convey information needed to conduct WP. We also develop a novel methodology for adaptively selecting the size of feature space, which optimizes over balancing the expected amount of useful information and classification capability, quantified using information-theoretic measures on the signal bin selection. Numerical results show that P-NN achieves a significant advantage in performance-complexity tradeoff over deep learning baselines that leverage the full power delay profile (PDP).

As Federated Learning (FL) grows in popularity, new decentralized frameworks are becoming widespread. These frameworks leverage the benefits of decentralized environments to enable fast and energy-efficient inter-device communication. However, this growing popularity also intensifies the need for robust security measures. While existing research has explored various aspects of FL security, the role of adversarial node placement in decentralized networks remains largely unexplored. This paper addresses this gap by analyzing the performance of decentralized FL for various adversarial placement strategies when adversaries can jointly coordinate their placement within a network. We establish two baseline strategies for placing adversarial node: random placement and network centrality-based placement. Building on this foundation, we propose a novel attack algorithm that prioritizes adversarial spread over adversarial centrality by maximizing the average network distance between adversaries. We show that the new attack algorithm significantly impacts key performance metrics such as testing accuracy, outperforming the baseline frameworks by between 9% and 66.5% for the considered setups. Our findings provide valuable insights into the vulnerabilities of decentralized FL systems, setting the stage for future research aimed at developing more secure and robust decentralized FL frameworks.

Collaborative federated learning (CFL) enables device cooperation in training shared machine learning models without reliance on a parameter server. However, the absence of a parameter server also impacts vulnerabilities associated with adversarial attacks, including privacy inference and Byzantine attacks. In this context, this paper introduces a novel CFL framework that enables each device to individually determine the subset of devices to transmit FL parameters to over the wireless network, based on its neighboring devices' location, current loss, and connection information, to achieve privacy protection and robust aggregation. This is formulated as an optimization problem whose goal is to minimize CFL training loss while satisfying the privacy preservation, robust aggregation, and transmission delay requirements. To solve this problem, a proximal policy optimization (PPO)-based reinforcement learning (RL) algorithm integrated with a graph neural network (GNN) is proposed. Compared to traditional algorithms that use global information with high computational complexity, the proposed GNN-RL method can be deployed on devices based on neighboring information with lower computational overhead. Simulation results show that the proposed algorithm can protect data privacy and increase identification accuracy by 15% compared to an algorithm in which devices are partially clustered for model aggregation.

Multimodal federated learning (FL) aims to enrich model training in FL settings where devices are collecting measurements across multiple modalities (e.g., sensors measuring pressure, motion, and other types of data). However, key challenges to multimodal FL remain unaddressed, particularly in heterogeneous network settings: (i) the set of modalities collected by each device will be diverse, and (ii) communication limitations prevent devices from uploading all their locally trained modality models to the server. In this paper, we propose Federated Multimodal Fusion learning with Selective modality communication (FedMFS), a new multimodal fusion FL methodology that can tackle the above mentioned challenges. The key idea is the introduction of a modality selection criterion for each device, which weighs (i) the impact of the modality, gauged by Shapley value analysis, against (ii) the modality model size as a gauge for communication overhead. This enables FedMFS to flexibly balance performance against communication costs, depending on resource constraints and application requirements. Experiments on the real-world ActionSense dataset demonstrate the ability of FedMFS to achieve comparable accuracy to several baselines while reducing the communication overhead by over 4x.

Federated learning (FL) encounters scalability challenges when implemented over fog networks. Semi-decentralized FL (SD-FL) proposes a solution that divides model cooperation into two stages: at the lower stage, device-to-device (D2D) communications is employed for local model aggregations within subnetworks (subnets), while the upper stage handles device-server (DS) communications for global model aggregations. However, existing SD-FL schemes are based on gradient diversity assumptions that become performance bottlenecks as data distributions become more heterogeneous. In this work, we develop semi-decentralized gradient tracking (SD-GT), the first SD-FL methodology that removes the need for such assumptions by incorporating tracking terms into device updates for each communication layer. Analytical characterization of SD-GT reveals convergence upper bounds for both non-convex and strongly-convex problems, for a suitable choice of step size. We employ the resulting bounds in the development of a co-optimization algorithm for optimizing subnet sampling rates and D2D rounds according to a performance-efficiency trade-off. Our subsequent numerical evaluations demonstrate that SD-GT obtains substantial improvements in trained model quality and communication cost relative to baselines in SD-FL and gradient tracking on several datasets.

Cognitive satellite-terrestrial networks (CSTNs) are a promising technology that optimize satellite efficiency and coverage. In this paper, we present a novel QoS-aware, deep learning (DL) algorithm that integrates with space-air-ground channels and exploits beam utilization by maximizing user admittance. To this end, we characterize the spectral radius as a critical concept that is indicative of resource utilization in the multibeam cluster. Our findings show that a smart use of the spectral radius, learned with ordinary differential equation networks (ODE-Nets), could maximize user admittance and outperform baselines like overlay and underlay, and predict the optimal power allocation vector.

While traditional federated learning (FL) typically focuses on a star topology where clients are directly connected to a central server, real-world distributed systems often exhibit hierarchical architectures. Hierarchical FL (HFL) has emerged as a promising solution to bridge this gap, leveraging aggregation points at multiple levels of the system. However, existing algorithms for HFL encounter challenges in dealing with multi-timescale model drift, i.e., model drift occurring across hierarchical levels of data heterogeneity. In this paper, we propose a multi-timescale gradient correction (MTGC) methodology to resolve this issue. Our key idea is to introduce distinct control variables to (i) correct the client gradient towards the group gradient, i.e., to reduce client model drift caused by local updates based on individual datasets, and (ii) correct the group gradient towards the global gradient, i.e., to reduce group model drift caused by FL over clients within the group. We analytically characterize the convergence behavior of MTGC under general non-convex settings, overcoming challenges associated with couplings between correction terms. We show that our convergence bound is immune to the extent of data heterogeneity, confirming the stability of the proposed algorithm against multi-level non-i.i.d. data. Through extensive experiments on various datasets and models, we validate the effectiveness of MTGC in diverse HFL settings. The code for this project is available at https://github.com/wenzhifang/MTGC.

Matrix computations are a fundamental building block of the edge computing systems, with a major recent uptick in demand due to their use in AI/ML training and inference procedures. Existing approaches for distributing the matrix computations involve allocating coded combinations of submatrices to worker nodes, to build resilience to slower nodes, called stragglers. In the edge learning context, however, these approaches will compromise sparsity properties that are often present in the original matrices found at the edge server. In this study, we consider the challenge of augmenting, such approaches to preserve input sparsity when distributing the task across the edge devices, thereby retaining the associated computational efficiency enhancements. First, we find a lower bound on the weight of coding, i.e., the number of submatrices to be combined to obtain coded submatrices to provide the resilience to the maximum possible number of straggler devices (for given number of devices and their storage constraints). Next, we propose distributed matrix computation schemes which meet the exact lower bound on the weight of the coding. Numerical experiments conducted in amazon Web services (AWSs) validate our assertions regarding straggler mitigation and computation speed for the sparse matrices.

Federated learning (FL) has been gaining attention for its ability to share knowledge while maintaining user data, protecting privacy, increasing learning efficiency, and reducing communication overhead. Decentralized FL (DFL) is a decentralized network architecture that eliminates the need for a central server in contrast to centralized FL (CFL). DFL enables direct communication between clients, resulting in significant savings in communication resources. In this paper, a comprehensive survey and profound perspective are provided for DFL. First, a review of the methodology, challenges, and variants of CFL is conducted, laying the background of DFL. Then, a systematic and detailed perspective on DFL is introduced, including iteration order, communication protocols, network topologies, paradigm proposals, and temporal variability. Next, based on the definition of DFL, several extended variants and categorizations are proposed with state-of-the-art (SOTA) technologies. Lastly, in addition to summarizing the current challenges in the DFL, some possible solutions and future research directions are also discussed.

Devices located in remote regions often lack coverage from well-developed terrestrial communication infrastructure. This not only prevents them from experiencing high quality communication services but also hinders the delivery of machine learning services in remote regions. In this paper, we propose a new federated learning (FL) methodology tailored to space-air-ground integrated networks (SAGINs) to tackle this issue. Our approach strategically leverages the nodes within space and air layers as both 1) edge computing units and 2) model aggregators during the FL process, addressing the challenges that arise from the limited computation powers of ground devices and the absence of terrestrial base stations in the target region. The key idea behind our methodology is the adaptive data offloading and handover procedures that incorporate various network dynamics in SAGINs, including the mobility, heterogeneous computation powers, and inconsistent coverage times of incoming satellites. We analyze the latency of our scheme and develop an adaptive data offloading optimizer, and also characterize the theoretical convergence bound of our proposed algorithm. Experimental results confirm the advantage of our SAGIN-assisted FL methodology in terms of training time and test accuracy compared with various baselines.

While network coverage maps continue to expand, many devices located in remote areas remain unconnected to terrestrial communication infrastructures, preventing them from getting access to the associated data-driven services. In this paper, we propose a ground-to-satellite cooperative federated learning (FL) methodology to facilitate machine learning service management over remote regions. Our methodology orchestrates satellite constellations to provide the following key functions during FL: (i) processing data offloaded from ground devices, (ii) aggregating models within device clusters, and (iii) relaying models/data to other satellites via inter-satellite links (ISLs). Due to the limited coverage time of each satellite over a particular remote area, we facilitate satellite transmission of trained models and acquired data to neighboring satellites via ISL, so that the incoming satellite can continue conducting FL for the region. We theoretically analyze the convergence behavior of our algorithm, and develop a training latency minimizer which optimizes over satellite-specific network resources, including the amount of data to be offloaded from ground devices to satellites and satellites’ computation speeds. Through experiments on three datasets, we show that our methodology can significantly speed up the convergence of FL compared with terrestrial-only and other satellite baseline approaches.

Radio access networks (RANs) in monolithic architectures have limited adaptability to supporting different network scenarios. Recently, open-RAN (O-RAN) techniques have begun adding enormous flexibility to RAN implementations. O-RAN is a natural architectural fit for cell-free massive multiple-input multiple-output (CFmMIMO) systems, where many geographically-distributed access points (APs) are employed to achieve ubiquitous coverage and enhanced user performance. In this paper, we address the decentralized pilot assignment (PA) problem for scalable O-RAN-based CFmMIMO systems. We propose a low-complexity PA scheme using a multi-agent deep reinforcement learning (MA-DRL) framework in which multiple learning agents perform distributed learning over the O-RAN communication architecture to suppress pilot contamination. Our approach does not require prior channel knowledge but instead relies on real-time interactions made with the environment during the learning procedure. In addition, we design a codebook search (CS) scheme that exploits the decentralization of our O-RAN CFmMIMO architecture, where different codebook sets can be utilized to further improve PA performance without any significant additional complexities. Numerical evaluations verify that our proposed scheme provides substantial computational scalability advantages and improvements in channel estimation performance compared to the state-of-the-art.

Recently, deep learning approaches have provided solutions to difficult problems in wireless positioning (WP). Although these WP algorithms have attained excellent and consistent performance against complex channel environments, the computational complexity coming from processing high-dimensional features can be prohibitive for mobile applications. In this work, we design a novel positioning neural network (P-NN) that utilizes the minimum description features to substantially reduce the complexity of deep learning-based WP. P-NN’s feature selection strategy is based on maximum power measurements and their temporal locations to convey information needed to conduct WP. We improve P-NN’s learning ability by intelligently processing two different types of inputs: sparse image and measurement matrices. Specifically, we implement a self-attention layer to reinforce the training ability of our network. We also develop a technique to adapt feature space size, optimizing over the expected information gain and the classification capability quantified with information-theoretic measures on signal bin selection. Numerical results show that P-NN achieves a significant advantage in performance-complexity tradeoff over deep learning baselines that leverage the full power delay profile (PDP). In particular, we find that P-NN achieves a large improvement in performance for low SNR, as unnecessary measurements are discarded in our minimum description features.

The demand for intelligent services at the network edge has introduced several research challenges. One is the need for a machine learning architecture that achieves personalization (to individual clients) and generalization (to unseen data) properties concurrently across different applications. Another is the need for an inference strategy that can satisfy network resource and latency constraints during testing-time. Existing techniques in federated learning have encountered a steep trade-off between personalization and generalization, and have not explicitly considered the resource requirements during the inference-stage. In this paper, we propose SplitGP, a joint edge-AI training and inference strategy that simultaneously captures generalization/personalization for efficient inference across resource-constrained clients. The training process of SplitGP is based on federated split learning, with the key idea of optimizing the client-side model to have personalization capability tailored to its main task, while training the server-side model to have generalization capability for handling out-of-distribution tasks. During testing-time, each client selectively offloads inference tasks to the server based on the uncertainty threshold tunable based on network resource availability. Through formal convergence analysis and inference time analysis, we provide guidelines on the selection of key meta-parameters in SplitGP. Experimental results confirm the advantage of SplitGP over existing baselines.

Federated learning (FedL) has emerged as a popular technique for distributing model training over a set of wireless devices, via iterative local updates (at devices) and global aggregations (at the server). In this paper, we develop parallel successive learning (PSL), which expands the FedL architecture along three dimensions: (i) Network, allowing decentralized cooperation among the devices via device-to-device (D2D) communications. (ii) Heterogeneity, interpreted at three levels: (ii-a) Learning: PSL considers heterogeneous number of stochastic gradient descent iterations with different mini-batch sizes at the devices; (ii-b) Data: PSL presumes a dynamic environment with data arrival and departure, where the distributions of local datasets evolve over time, captured via a new metric for model/concept drift. (ii-c) Device: PSL considers devices with different computation and communication capabilities. (iii) Proximity, where devices have different distances to each other and the access point. PSL considers the realistic scenario where global aggregations are conducted with idle times in-between them for resource efficiency improvements, and incorporates data dispersion and model dispersion with local model condensation into FedL. Our analysis sheds light on the notion of cold vs. warmed up models, and model inertia in distributed machine learning. We then propose network-aware dynamic model tracking to optimize the model learning vs. resource efficiency tradeoff, which we show is an NP-hard signomial programming problem. We finally solve this problem through proposing a general optimization solver. Our numerical results reveal new findings on the interdependencies between the idle times in-between the global aggregations, model/concept drift, and D2D cooperation configuration.

The conventional federated learning (FedL) architecture distributes machine learning (ML) across worker devices by having them train local models that are periodically aggregated by a server. FedL ignores two important characteristics of contemporary wireless networks, however: (i) the network may contain heterogeneous communication/computation resources, and (ii) there may be significant overlaps in devices’ local data distributions. In this work, we develop a novel optimization methodology that jointly accounts for these factors via intelligent device sampling complemented by device-to-device (D2D) offloading. Our optimization methodology aims to select the best combination of sampled nodes and data offloading configuration to maximize FedL training accuracy while minimizing data processing and D2D communication resource consumption subject to realistic constraints on the network topology and device capabilities. Theoretical analysis of the D2D offloading subproblem leads to new FedL convergence bounds and an efficient sequential convex optimizer. Using these results, we develop a sampling methodology based on graph convolutional networks (GCNs) which learns the relationship between network attributes, sampled nodes, and D2D data offloading to maximize FedL accuracy. Through evaluation on popular datasets and real-world network measurements from our edge testbed, we find that our methodology outperforms popular device sampling methodologies from literature in terms of ML model performance, data processing overhead, and energy consumption.

This paper considers improving wireless communication and computation efficiency in federated learning (FL) via model quantization. In the proposed bitwidth FL scheme, edge devices train and transmit quantized versions of their local FL model parameters to a coordinating server, which, in turn, aggregates them into a quantized global model and synchronizes the devices. The goal is to jointly determine the bitwidths employed for local FL model quantization and the set of devices participating in FL training at each iteration. We pose this as an optimization problem that aims to minimize the training loss of quantized FL under a per-iteration device sampling budget and delay requirement. However, the formulated problem is difficult to solve without (i) a concrete understanding of how quantization impacts global ML performance and (ii) the ability of the server to construct estimates of this process efficiently. To address the first challenge, we analytically characterize how limited wireless resources and induced quantization errors affect the performance of the proposed FL method. Our results quantify how the improvement of FL training loss between two consecutive iterations depends on the device selection and quantization scheme as well as on several parameters inherent to the model being learned. Then, to address the second challenge, we show that the FL training process can be described as a Markov decision process (MDP) and propose a model-based reinforcement learning (RL) method to optimize action selection over iterations. Compared to model-free RL, this model-based RL approach leverages the derived mathematical characterization of the FL training process to discover an effective device selection and quantization scheme without imposing additional device communication overhead. Simulation results show that the proposed FL algorithm can reduce the convergence time by 29% and 63% compared to a model free RL method and the standard FL method, respectively.

In this paper, the performance optimization of federated learning (FL), when deployed over a realistic wireless multiple-input multiple-output (MIMO) communication system with digital modulation and over-the-air computation (AirComp) is studied. In particular, a MIMO system is considered in which edge devices transmit their local FL models (trained using their locally collected data) to a parameter server (PS) using beamforming to maximize the number of devices scheduled for transmission. The PS, acting as a central controller, generates a global FL model using the received local FL models and broadcasts it back to all devices. Due to the limited bandwidth in a wireless network, AirComp is adopted to enable efficient wireless data aggregation. However, fading of wireless channels can produce aggregate distortions in an AirComp-based FL scheme. To tackle this challenge, we propose a modified federated averaging (FedAvg) algorithm that combines digital modulation with AirComp to mitigate wireless fading while ensuring the communication efficiency. This is achieved by a joint transmit and receive beamforming design, which is formulated as an optimization problem to dynamically adjust the beamforming matrices based on current FL model parameters so as to minimize the transmitting error and ensure the FL performance. To achieve this goal, we first analytically characterize how the beamforming matrices affect the performance of the FedAvg in different iterations. Based on this relationship, an artificial neural network (ANN) is used to estimate the local FL models of all devices and adjust the beamforming matrices at the PS for future model transmission. The algorithmic advantages and improved performance of the proposed methodologies are demonstrated through extensive numerical experiments.

We consider error-correction coding schemes for adversarial wiretap channels (AWTCs) in which the channel can a) read a fraction of the codeword bits up to a bound r and b) flip a fraction of the bits up to a bound p. The channel can freely choose the locations of the bit reads and bit flips via a process with unbounded computational power. Codes for the AWTC are of broad interest in the area of information security, as they can provide data resiliency in settings where an attacker has limited access to a storage or transmission medium. We investigate a family of non-linear codes known as pseudolinear codes, which were first proposed by Guruswami and Indyk (FOCS 2001) for constructing list-decodable codes independent of the AWTC setting. Unlike general non-linear codes, pseudolinear codes admit efficient encoders and have succinct representations. We focus on unique decoding and show that random pseudolinear codes can achieve rates up to the binary symmetric channel (BSC) capacity $1-H_{2}(p)$ for any $p, r$ in the less noisy region: $p\lt1/2$ and $r\lt1-H_{2}(p)$ where $H_{2}(\cdot)$ is the binary entropy function. Thus, pseudolinear codes are the first known optimal-rate binary code family for the less noisy AWTC that admit efficient encoders. The above result can be viewed as a derandomization result of random general codes in the AWTC setting, which in turn opens new avenues for applying derandomization techniques to randomized constructions of AWTC codes. Our proof applies a novel concentration inequality for sums of random variables with limited independence which may be of interest as an analysis tool more generally.

Augmenting federated learning (FL) with direct device-to-device (D2D) communications can help improve conver-gence speed and reduce model bias through rapid local information exchange. However, data privacy concerns, device trust issues, and unreliable wireless channels each pose challenges to determining an effective yet resource efficient D2D structure. In this paper, we develop a decentralized reinforcement learning (RL) methodology for D2D graph discovery that promotes communication of non-sensitive yet impactful data-points over trusted yet reliable links. Each device functions as an RL agent, training a policy to predict the impact of incoming links. Local (device-level) and global rewards are coupled through message passing within and between device clusters. Numerical experiments confirm the advantages offered by our method in terms of convergence speed and straggler resilience across several datasets and FL schemes.

In this paper, we consider the design of an energy efficient collaborative federated learning (CFL) methodology where devices exchange their local FL parameters with a subset of their neighbors without reliance on a parameter server. In the considered model, mobile devices implement the designed CFL to train their local FL models using their own datasets over a realistic wireless network. Due to the limited wireless resources and user movements, each device may not be able to transmit its FL parameters with all neighboring devices. Therefore, each device must select a subset of devices to share its FL parameters and optimize the transmit power. This problem is formulated as an optimization problem, whose goal is to minimize CFL training energy consumption while satisfying the delay and CFL training loss requirements. To solve this problem, a two-stage solution is proposed. At the first stage, a graph neural network (GNN) based algorithm is proposed, which enables each device to individually determine the subset of devices to transmit FL parameters using its neighboring devices' location and connection information. Compared to standard iterative algorithms that need to iteratively optimize device connections and transmit power, the proposed GNN based method can directly obtain the optimal device connections without iterative optimization. Given the optimal device connections, at the second stage, each device can directly obtain the optimal transmit power. Simulation results show that the proposed algorithm can decrease energy consumption by up to 46% compared to the algorithm where each device will directly connect to its first and second nearest neighbors.

In this paper, we consider the optimization of federated learning (FL) over a realistic wireless multiple-input multiple-output (MIMO) communication system with digital modulation and over-the-air computation (AirComp). In such a system, MIMO devices transmit their locally trained FL models to a parameter server (PS) using beamforming to maximize the number of devices scheduled for transmission. AirComp enables efficient wireless model aggregation by the PS in bandwidth-limited settings. However, wireless channel fading can produce distortions in AirComp-based FL. To tackle this challenge, we develop a novel aggregation scheme that combines digital modulation with AirComp to mitigate wireless fading while ensuring communication efficiency. We formulate this as a joint transmit-receive beamforming design optimization problem which dynamically adjusts the beamforming matrices to minimize the FL training loss with transmission errors. To solve this problem based on limited information at the PS, we employ an artificial neural network (ANN) to estimate the local FL models of all devices. Then, we derive a closed-form optimal design of the transmit and receive beamforming matrices based on predicted FL models. Numerical evaluations validate the advantages of the proposed methodology in terms of model training performance compared with baselines.

Federated learning (FL) is a popular technique for training a global model on data distributed across client devices. Like other distributed training techniques, FL is susceptible to straggler (slower or failed) clients. Recent work has proposed to address this through device-to-device (D2D) offloading, which introduces privacy concerns. In this paper, we propose a novel straggler-optimal approach for coded matrix computations which can significantly reduce the communication delay and privacy issues introduced from D2D data transmissions in FL. Moreover, our proposed approach leads to a considerable improvement of the local computation speed when the generated data matrix is sparse. Numerical evaluations confirm the superiority of our proposed method over baseline approaches.

Opportunistic spectrum access is a viable technique for cognitive radio (CR) networks to address the spectrum scarcity problem, where both spectrum sensing and resource allocation (SSRA) are significant to the system throughput performance. Previous works on SSRA often require complete network statistics which may not be feasible given the time-varying nature of practical CR networks. In this paper, we propose a learning-based optimization framework for SSRA in multi-band-multi-user CR networks. We develop a dynamic cooperative spectrum sensing strategy which allows secondary users to detect available spectrum bands of the primary user, followed by flexible power allocation for efficient data transmissions. To cope with the dynamic of channel and resource statistics, we propose an improved deep reinforcement learning scheme based on a maximum entropy-enabled actor critic algorithm. Numerical results demonstrate the superiority of our approach over existing schemes.

There has been recent interest in leveraging federated learning (FL) for radio signal classification tasks. In FL, model parameters are periodically communicated from participating devices, training on their own local datasets, to a central server which aggregates them into a global model. While FL has privacy/security advantages due to raw data not leaving the devices, it is still susceptible to several adversarial attacks. In this work, we reveal the susceptibility of FL-based signal classifiers to model poisoning attacks, which compromise the training process despite not observing data transmissions. In this capacity, we develop an attack framework in which compromised FL devices perturb their local datasets using adversarial evasion attacks. As a result, the training process of the global model significantly degrades on in-distribution signals (i.e., signals received over channels with identical distributions at each edge device). We compare our work to previously proposed FL attacks and reveal that as few as one adversarial device operating with a low-powered perturbation under our attack framework can induce the potent model poisoning attack to the global classifier. Moreover, we find that more devices partaking in adversarial poisoning will proportionally degrade the classification performance.

The two leading solution paradigms for imitation learning (IL), BC and GAIL, each suffers from notable drawbacks. BC, a supervised learning approach to mimic expert actions, is vulnerable to covariate shift. GAIL applies adversarial training to minimize the discrepancy between expert and learner behaviors, which is prone to unstable training and mode collapse. In this work, we propose DC – Distributional Cloning – a novel IL approach for addressing the covariate shift and mode collapse problems simultaneously. DC directly maximizes the likelihood of observed expert and learner demonstrations, and gradually encourages the learner to evolve towards expert behaviors based on an averaging effect. The DC solution framework contains two stages in each training loop, where in stage one the mixed expert and learner state distribution is estimated via SoftFlow, and in stage two the learner policy is trained to match both the expert’s policy and state distribution via ADMM. Experimental evaluation of DC compared with several baselines in 10 different physics-based control tasks reveal superior results in learner policy performance, training stability, and mode distribution preservation.

The design of codes for feedback-enabled communications has been a long-standing open problem. Recent research on non-linear, deep learning-based coding schemes have demonstrated significant improvements in communication reliability over linear codes, but are still vulnerable to the presence of forward and feedback noise over the channel. In this paper, we develop a new family of non-linear feedback codes that greatly enhance robustness to channel noise. Our autoencoder-based architecture is designed to learn codes based on consecutive blocks of bits, which obtains de-noising advantages over bit-by-bit processing to help overcome the physical separation between the encoder and decoder over a noisy channel. Moreover, we develop a power control layer at the encoder to explicitly incorporate hardware constraints into the learning optimization, and prove that the resulting average power constraint is satisfied asymptotically. Numerical experiments demonstrate that our scheme outperforms state-of-the-art feedback codes by wide margins over practical forward and feedback noise regimes, and provide information-theoretic insights on the behavior of our non-linear codes. Moreover, we observe that, in a long blocklength regime, canonical error correction codes are still preferable to feedback codes when the feedback noise becomes high. Our code is available at https://anonymous.4open.science/r/RCode1.

A fundamental challenge to providing edge-AI services is the need for a machine learning (ML) model that achieves personalization (i.e., to individual clients) and generalization (i.e., to unseen data) properties concurrently. Existing techniques in federated learning (FL) have encountered a steep tradeoff between these objectives and impose large computational requirements on edge devices during training and inference. In this paper, we propose SplitGP, a new split learning solution that can simultaneously capture generalization and personalization capabilities for efficient inference across resource-constrained clients (e.g., mobile/IoT devices). Our key idea is to split the full ML model into client-side and server-side components, and impose different roles to them: the client-side model is trained to have strong personalization capability optimized to each client’s main task, while the server-side model is trained to have strong generalization capability for handling all clients’ out-of-distribution tasks. We analytically characterize the convergence behavior of SplitGP, revealing that all client models approach stationary points asymptotically. Further, we analyze the inference time in SplitGP and provide bounds for determining model split ratios. Experimental results show that SplitGP outperforms existing baselines by wide margins in inference time and test accuracy for varying amounts of out-of-distribution samples.

Straggler nodes are well-known bottlenecks of distributed matrix computations which induce reductions in computation/communication speeds. A common strategy for mitigating such stragglers is to incorporate MDS (maximum distance separable) codes into the framework; this can achieve resilience against an optimal number of stragglers. However, these codes assign dense linear combinations of submatrices to the workers which increase the number of non-zero entries in the encoded matrices, and adversely affect the worker computation time. In this work, we develop a straggler-optimal distributed matrix computation approach where the assigned encoded submatrices are linear combinations of a small number of submatrices so that it is well suited for sparse input matrices. Numerical experiments conducted in Amazon Web Services (AWS) demonstrate up to 30% reduction in worker computation time and 100 times faster encoding compared to several recent methods.

Semi-decentralized federated learning blends the conventional device-to-server (D2S) interaction structure of federated model training with localized device-to-device (D2D) communications. We study this architecture over practical edge networks with multiple D2D clusters modeled as time-varying and directed communication graphs. Our investigation results in an algorithm that controls the fundamental trade-off between (a) the rate of convergence of the model training process towards the global optimizer, and (b) the number of D2S transmissions required for global aggregation. Specifically, in our semi-decentralized methodology, D2D consensus updates are injected into the federated averaging framework based on column-stochastic weight matrices that encapsulate the connectivity within the clusters. To arrive at our algorithm, we show how the expected optimality gap in the current global model depends on the greatest two singular values of the weighted adjacency matrices (and hence on the densities) of the D2D clusters. We then derive tight bounds on these singular values in terms of the node degrees of the D2D clusters, and we use the resulting expressions to design a threshold on the number of clients required to participate in any given global aggregation round so as to ensure a desired convergence rate. Simulations performed on real-world datasets reveal that our connectivity-aware algorithm reduces the total communication cost required to reach a target accuracy significantly compared with baselines depending on the connectivity structure and the learning task.

Federated reinforcement learning (FedRL) enables agents to collaboratively train a global policy without sharing their individual data. However, high communication overhead remains a critical bottleneck, particularly for natural policy gradient (NPG) methods, which are second-order. To address this issue, we propose the FedNPG-ADMM framework, which leverages the alternating direction method of multipliers (ADMM) to approximate global NPG directions efficiently. We theoretically demonstrate that using ADMM-based gradient updates reduces communication complexity from $\mathcal{O}({d^{2}})$ to $\mathcal{O}({d})$ at each iteration, where $d$ is the number of model parameters. Furthermore, we show that achieving an $\epsilon$-error stationary convergence requires $\mathcal{O}(\frac{1}{(1-\gamma)^{2}{\epsilon}})$ iterations for discount factor $\gamma$, demonstrating that FedNPG-ADMM maintains the same convergence rate as standard FedNPG. Through evaluation of the proposed algorithms in MuJoCo environments, we demonstrate that FedNPG-ADMM maintains the reward performance of standard FedNPG, and that its convergence rate improves when the number of federated agents increases.

Traditional federated learning (FL) algorithms operate under the assumption that the data distributions at training (source domains) and testing (target domain) are the same. The fact that domain shifts often occur in practice necessitates equipping FL methods with a domain generalization (DG) capability. However, existing DG algorithms face fundamental challenges in FL setups due to the lack of samples/domains in each client’s local dataset. In this paper, we propose StableFDG, a style and attention based learning strategy for accomplishing federated domain generalization, introducing two key contributions. The first is style-based learning, which enables each client to explore novel styles beyond the original source domains in its local dataset, improving domain diversity based on the proposed style sharing, shifting, and exploration strategies. Our second contribution is an attention-based feature highlighter, which captures the similarities between the features of data samples in the same class, and emphasizes the important/common characteristics to better learn the domain-invariant characteristics of each class in data-poor FL scenarios. Experimental results show that StableFDG outperforms existing baselines on various DG benchmark datasets, demonstrating its efficacy.

Federated machine learning (FL) is a powerful technology which can be implemented to exploit the sheer amount of geographically distributed data for enhanced computation. Exploiting the impending proliferation of wireless devices, in this paper, we incorporate distributed quantized transmissions for reliable connectivity to a remote FL server. We develop a novel theoretical framework for the convergence analysis of the proposed network under joint impact of communication bit error rate (BER), and model quantization, and participation control. We show that the convergence rate of the network is affected by the BER and it can be improved via participation control. Through simulation, we demonstrate that our proposed model can provide the same performance as the conventional FL networks based on point-to-point communication while the energy consumption is divided across the distributed nodes.

Lightweight neural networks refer to deep networks with small numbers of parameters, which can be deployed in resource-limited hardware such as embedded systems. To learn such lightweight networks effectively and efficiently, in this paper we propose a novel convolutional layer, namely Channel-Split Recurrent Convolution (CSR-Conv), where we split the output channels to generate data sequences with length T as the input to the recurrent layers with shared weights. As a consequence, we can construct lightweight convolutional networks by simply replacing (some) linear convolutional layers with CSR-Conv layers. We prove that under mild conditions the model size decreases with the rate of $O\left( {\frac{1}{{{T^2}}}} \right)$. Empirically we demonstrate the state-of-the-art performance using VGG-16, ResNet-50, ResNet-56, ResNet-110, DenseNet-40, MobileNet, and EfficientNet as backbone networks on CIFAR-10 and ImageNet. Codes can be found on https://github.com/tuaxon/CSR_Conv.

Low latency, resource efficiency, and data privacy are some of the crucial requirements in modern communication networks. Federated learning can efficiently address these issues by utilizing the data at the network edge and processing massive amounts of data in parallel at the edge devices, thus ensuring data privacy and low latency. For a large-scale federated learning task spanning many devices, challenges arise due to device heterogeneity, data variability, and limited network resources. Optimal selection of edge devices participating in federated learning is essential to attaining resilient, reliable, and resource-efficient edge networks. In this context, this article proposes an optimal device selection method to minimize redundant data training and improve network resource utilization without affecting the performance of federated learning over resource-constrained edge networks. The proposed optimal device selection method aims to minimize network resource demands while maximizing data diversity within the aggregated model. The performance of the proposed federated learning framework is evaluated using a publicly available image data set of handwritten digits, EMNIST (an extended version of the MNIST data set). Experimental results indicate that the proposed framework can obtain accuracy convergence performance on par with conventional federated learning methods while significantly reducing device usage (up to$50\%$) and resource utilization (up to$30\%$) while reaching 99% of achievable accuracy. The proposed method can therefore be effectively applied to resource-constrained edge networks.

This paper introduces AGAPECert, an Auditable, Generalized, Automated, Privacy-Enabling, Certification framework capable of performing auditable computation on private data and reporting real-time aggregate certification status without disclosing underlying private data. AGAPECert utilizes a novel mix of trusted execution environments, blockchain technologies, and a real-time graph-based API standard to provide automated, oblivious, and auditable certification. Our technique allows a privacy-conscious data owner to run pre-approvedOblivious Smart Contractcode in their own environment on their own private data to produce Private Automated Certifications. These certifications are verifiable, purely functional transformations of the available data, enabling a third party to trust that the private data must have the necessary properties to produce the resulting certification. Recently, a multitude of solutions for certification and traceability in supply chains have been proposed. These often suffer from significant privacy issues because they tend to take a ”shared, replicated database” approach: every node in the network has access to a copy of all relevant data and contract code to guarantee the integrity and reach consensus, even in the presence of malicious nodes. In these contexts of certifications that require global coordination, AGAPECert can include a blockchain to guarantee ordering of events, while keeping a core privacy model where private data is not shared outside of the data owner's own platform. AGAPECert contributes an open-source certification framework that can be adopted in any regulated environment to keep sensitive data private while enabling a trusted automated workflow.

The proliferation of data-intensive mobile applications is causing latency to become an issue in mobile edge computing (MEC) systems. In this work, we propose a novel methodology that optimizes communication, computation, and caching configurations in MEC to minimize the mean latency experienced by mobile devices. Transmission and computation processes are modeled using M/G/1 queues to account for service rates and warm-up times. Our caching scheme includes time variables for each file at each edge server in determining when to discard files from storage. We theoretically analyze the latency experienced by mobile devices due to communication, computation, and caching, showing how MEC system latency depends on the offloading decisions of mobile devices, bandwidth and CPU resources, and expiration times of files in the storage of edge servers. Our method for solving the latency minimization problem consists of two main components: iNner cOnVex Approximation (NOVA) to deal with non-convexity in the optimization, and an online algorithm for preventing cache storage violations as new tasks arrive and are serviced by the MEC system. Simulation results show that our algorithm outperforms several baselines in minimizing latency, and verify the benefit of including different resource allocation variables in our optimization.

We investigate training machine learning (ML) models across a set of geo-distributed, resource-constrained clusters of devices through unmanned aerial vehicles (UAV) swarms. The presence of time-varying data heterogeneity and computational resource inadequacy among device clusters motivate four key parts of our methodology: (i)stratified UAV swarmsof leader, worker, and coordinator UAVs, (ii)hierarchical nested personalized federated learning(HN-PFL), a distributed ML framework for personalized model training across the worker-leader-core network hierarchy, (iii)cooperative UAV resource poolingto address computational inadequacy of devices by conducting model training among the UAV swarms, and (iv)model/concept driftto model time-varying data distributions. In doing so, we consider bothmicro(i.e., UAV-level) andmacro(i.e., swarm-level) system design. At the micro-level, we propose network-awareHN-PFL, where we distributively orchestrate UAVs inside swarms to optimize energy consumption and ML model performance with performance guarantees. At the macro-level, we focus on swarm trajectory and learning duration design, which we formulate as a sequential decision making problem tackled via deep reinforcement learning. Our simulations demonstrate the improvements achieved by our methodology in terms of ML performance, network resource savings, and swarm trajectory efficiency.

We propose cooperative edge-assisted dynamic federated learning (CE-FL).CE-FLintroduces a distributed machine learning (ML) architecture, where data collection is carried out at the end devices, while the model training is conducted cooperatively at the end devices and the edge servers, enabled via data offloading from the end devices to the edge servers through base stations.CE-FLalso introduces floating aggregation point, where the local models generated at the devices and the servers are aggregated at an edge server, which varies from one model training round to another to cope with the network evolution in terms of data distribution and users’ mobility.CE-FLconsiders the heterogeneity of network elements in terms of communication/computation models and the proximity to one another.CE-FLfurther presumes a dynamic environment with online variation of data at the network devices which causes a drift at the ML model performance. We model the processes taken duringCE-FL, and conduct analytical convergence analysis of its ML model training. We then formulate network-awareCE-FLwhich aims to adaptively optimize all the network elements via tuning their contribution to the learning process, which turns out to be a non-convex mixed integer problem. Motivated by the large scale of the system, we propose a distributed optimization solver to break down the computation of the solution across the network elements. We finally demonstrate the effectiveness of our framework with the data collected from a real-world testbed.

We consider the problem of predicting link formation in Social Learning Networks (SLN), a type of social network that forms when people learn from one another through structured interactions. While link prediction has been studied for general types of social networks, the evolution of SLNs over their lifetimes coupled with their dependence on which topics are being discussed presents new challenges for this type of network. To address these challenges, we develop a series of autonomous link prediction methodologies that utilize spatial and time-evolving network architectures to pass network state between space and time periods, and that models over three types of SLN features updated in each period: neighborhood-based (e.g., resource allocation), path-based (e.g., shortest path), and post-based (e.g., topic similarity). Through evaluation on six real-world datasets from Massive Open Online Course (MOOC) discussion forums and from Purdue University, we find that our method obtains substantial improvements over Bayesian models, linear classifiers, and graph neural networks, with AUCs typically above 0.91 and reaching 0.99 depending on the dataset. Our feature importance analysis shows that while neighborhood and path-based features contribute the most to the results, post-based features add additional information that may not always be relevant for link prediction. The code and four of the datasets used in this work are available athttps://github.com/Jess-jpg-txt/sln-learning.

We study the problem of learning representations that are private yet informative i.e., provide information about intended "ally" targets while hiding sensitive "adversary" attributes. We propose Exclusion-Inclusion Generative Adversarial Network (EIGAN), a generalized private representation learning (PRL) architecture that accounts for multiple ally and adversary attributes unlike existing PRL solutions. While centrally-aggregated dataset is a prerequisite for most PRL techniques, data in real-world is often siloed across multiple distributed nodes unwilling to share the raw data because of privacy concerns. We address this practical constraint by developing D-EIGAN, the first distributed PRL method that learns representations at each node without transmitting the source data. We theoretically analyze the behavior of adversaries under the optimal EIGAN and D-EIGAN encoders and the impact of dependencies among ally and adversary tasks on the optimization objective. Our experiments on various datasets demonstrate the advantages of EIGAN in terms of performance, robustness, and scalability. In particular, EIGAN outperforms the previous state-of-the-art by a significant accuracy margin ($47%$ improvement), and D-EIGAN’s performance is consistently on par with EIGAN under different network settings.

Traditional learning-based approaches to student modeling generalize poorly to underrepresented student groups due to biases in data availability. In this paper, we propose a methodology for predicting student performance from their online learning activities that optimizes inference accuracy over different demographic groups such as race and gender. Building upon recent foundations in federated learning, in our approach, personalized models for individual student subgroups are derived from a global model aggregated across all student models via meta-gradient updates that account for subgroup heterogeneity. To learn better representations of student activity, we augment our approach with a self-supervised behavioral pretraining methodology that leverages multiple modalities of student behavior (e.g., visits to lecture videos and participation on forums), and include a neural network attention mechanism in the model aggregation stage. Through experiments on three real-world datasets from online courses, we demonstrate that our approach obtains substantial improvements over existing student modeling baselines in predicting student learning outcomes for all subgroups. Visual analysis of the resulting student embeddings confirm that our personalization methodology indeed identifies different activity patterns within different subgroups, consistent with its stronger inference ability compared with the baselines.

Federated learning (FL) has been recognized as one of the most promising solutions for distributed machine learning (ML). In most of the current literature, FL has been studied for supervised ML tasks, in which edge devices collect labeled data. Nevertheless, in many applications, it is impractical to assume existence of labeled data across devices. To this end, we develop a novel methodology, Cooperative Federated unsupervised Contrastive Learning (CF-CL), for FL across edge devices with unlabeled datasets. CF-CL employs local device cooperation where data are exchanged among devices through device-to-device (D2D) communications to avoid local model bias resulting from non-independent and identically distributed (non-i.i.d.) local datasets. CF-CL introduces a push-pull smart data sharing mechanism tailored to unsupervised FL settings, in which, each device pushes a subset of its local datapoints to its neighbors as reserved datapoints, and pulls a set of datapoints from its neighbors, sampled through a probabilistic importance sampling technique. We demonstrate that CF-CL leads to (i) alignment of unsupervised learned latent spaces across devices, (ii) faster global convergence, allowing for less frequent global model aggregations; and (iii) is effective in extreme non-i.i.d. datasettings across the devices.

This paper considers improving wireless communication and computation efficiency in federated learning (FL) via model quantization. In the proposed bitwidth FL scheme, edge devices train and transmit quantized versions of their local FL model parameters to a coordinating server, which, in turn, aggregates them into a quantized global model and synchronizes the devices. With the goal of jointly determining the set of participating devices in each training iteration and the bitwidths employed at the devices, we pose an optimization problem for minimizing the training loss of quantized FL under a device sampling budget and delay requirement. Our analytical results show that the improvement of FL training loss between two consecutive iterations depends on not only the device selection and quantization scheme, but also on several parameters inherent to the model being learned. As a result, we propose, a model-based reinforcement learning (RL) method to optimize action selection over iterations. Compared to model-free RL, the proposed approach leverages the derived mathematical characterization of the FL training process to discover an effective device selection and quantization scheme without imposing additional device communication overhead. Numerical evaluations show that the proposed FL framework can achieve the same classification performance while reducing the number of training iterations needed for convergence by 20% compared to model-free RL-based FL.

Intelligent reflecting surfaces (IRS) consist of configurable meta-atoms, which can alter the wireless propagation environment through design of their reflection coefficients. We consider adaptive IRS control in the practical setting where (i) the IRS reflection coefficients are attained by adjusting tunable elements embedded in the meta-atoms, (ii) the IRS reflection coefficients are affected by the incident angles of the incoming signals, (iii) the IRS is deployed in multi-path, time-varying channels, and (iv) the feedback link from the base station (BS) to the IRS has a low data rate. Conventional optimization-based IRS control protocols, which rely on channel estimation and conveying the optimized variables to the IRS, are not practical in this setting due to the difficulty of channel estimation and the low data rate of the feedback channel. To address these challenges, we develop a novel adaptive codebook-based limited feedback protocol to control the IRS. We propose two solutions for adaptive IRS codebook design: (i) random adjacency (RA), which utilizes correlations across the channel realizations, and (ii) deep neural network policy-based IRS control (DPIC), which is based on a deep reinforcement learning. Numerical evaluations show that the data rate and average data rate over one coherence time are improved substantially by the proposed schemes.

Feedback is a critical element of student-instructor interaction: it provides a direct manner for students to learn from mistakes. However, with student to teacher ratios growing rapidly, challenges arise for instructors to provide quality feedback to individual students. While significant efforts have been directed at automating feedback generation, relatively little attention has been given to underlying feedback characteristics. We develop a methodology for analyzing instructor-provided feedback and determining how it correlates with changes in student grades using data from online higher education engineering classrooms. Specifically, we featurize written feedback on individual assignments using Natural Language Processing (NLP) techniques including sentiment analysis, bigram splitting, and Named Entity Recognition (NER) to quantify post-, sentence-, and word-dependent attributes of grader writing. We demonstrate that student grade improvement can be well approximated by a multivariate linear model with average fits across course sections between 67% and 83%. We determine several statistically significant contributors to and detractors from student success contained in instructor feedback. For example, our results reveal that inclusion of student name is significantly correlated with an improvement in post-feedback grades, as is inclusion of specific assignment-related keywords. Finally, we discuss how this methodology can be incorporated into educational technology systems to make recommendations for feedback content from observed student behavior.

Robust and accurate unmanned aircraft system (UAS) detection is pivotal in restricted air spaces. Deep learning-based object detection has been proposed to identify the presence of UASs, but it introduces two key challenges. Specifically, deep learning detectors (i) provide point estimates at test-time with no associated measure of uncertainty, and (ii) easily trigger false positive detections for birds and other aerial wildlife. In this work, we propose a novel detection algorithm, which is capable of providing uncertainty quantification (UQ) metrics at test time while also significantly reducing the false positive rate on natural wildlife. Our proposed method consists of using an ensemble of object detectors to generate a distributive estimate of each input prediction. In addition, we measure multiple UQ-based scoring metrics for each input to further validate our model’s effectiveness. Through evaluation on our custom generated UAS dataset, consisting of images captured from deployed cameras, we show that our model provides robust UQ estimates, low false positive rates on wildlife, and significantly improved error rates over singular deep learning detection models.

In this paper, we study a new latency optimization problem for blockchain-based federated learning (BFL) in multi-server edge computing. In this system model, distributed mobile devices (MDs) communicate with a set of edge servers (ESs) to handle both machine learning (ML) model training and block mining simultaneously. To assist the ML model training for resource-constrained MDs, we develop an offloading strategy that enables MDs to transmit their data to one of the associated ESs. We then propose a new decentralized ML model aggregation solution at the edge layer based on a consensus mechanism to build a global ML model via peer-to-peer (P2P)-based blockchain communications. Blockchain builds trust among MDs and ESs to facilitate reliable ML model sharing and cooperative consensus formation, and enables rapid elimination of manipulated models caused by poisoning attacks. We formulate latency-aware BFL as an optimization aiming to minimize the system latency via joint consideration of the data offloading decisions, MDs’ transmit power, channel bandwidth allocation for MDs’ data offloading, MDs’ computational allocation, and hash power allocation. Given the mixed action space of discrete offloading and continuous allocation variables, we propose a novel deep reinforcement learning scheme with a parameterized advantage actor critic algorithm. We theoretically characterize the convergence properties of BFL in terms of the aggregation delay, mini-batch size, and number of P2P communication rounds. Our numerical evaluation demonstrates the superiority of our proposed scheme over baselines in terms of model training efficiency, convergence rate, system latency, and robustness against model poisoning attacks.

The combinatorial auction (CA) is an efficient mechanism for resource allocation in different fields, including cloud computing. It can obtain high economic efficiency and user flexibility by allowing bidders to submit bids for combinations of different items instead of only for individual items. However, the problem of allocating items among the bidders to maximize the auctioneers’ revenue, i.e., the winner determination problem (WDP), is NP-complete to solve and inapproximable. Existing works for WDPs are generally based on mathematical optimization techniques and most of them focus on the single-unit WDP, where each item only has one unit. On the contrary, few works consider the multi-unit WDP in which each item may have multiple units. Given that the multi-unit WDP is more complicated but prevalent in cloud computing, we propose leveraging machine learning (ML) techniques to develop a novel low-complexity algorithm for solving this problem with negligible revenue loss. Specifically, we model the multi-unit WDP as an augmented bipartite bid-item graph and use a graph neural network (GNN) with half-convolution operations to learn the probability of each bid belonging to the optimal allocation. To improve the sample generation efficiency and decrease the number of needed labeled instances, we propose two different sample generation processes. We also develop two novel graph-based post-processing algorithms to transform the outputs of the GNN into feasible solutions. Through simulations on both synthetic instances and a specific virtual machine (VM) allocation problem in a cloud computing platform, we validate that our proposed method can approach optimal performance with low complexity and has good generalization ability in terms of problem size and user-type distribution.

Federated learning has generated significant interest, with nearly all works focused on a “star” topology where nodes/devices are each connected to a central server. We migrate away from this architecture and extend it through thenetworkdimension to the case where there are multiple layers of nodes between the end devices and the server. Specifically, we develop multi-stage hybrid federated learning (MH-FL), a hybrid of intra-and inter-layer model learning that considers the network as amulti-layer cluster-based structure.MH-FLconsiders thetopology structuresamong the nodes in the clusters, including local networks formed via device-to-device (D2D) communications, and presumes asemi-decentralized architecturefor federated learning. It orchestrates the devices at different network layers in a collaborative/cooperative manner (i.e., using D2D interactions) to formlocal consensuson the model parameters and combines it with multi-stage parameter relaying between layers of the tree-shaped hierarchy. We derive the upper bound of convergence forMH-FLwith respect to parameters of the network topology (e.g., the spectral radius) and the learning algorithm (e.g., the number of D2D rounds in different clusters). We obtain a set of policies for the D2D rounds at different clusters to guarantee either a finite optimality gap or convergence to the global optimum. We then develop a distributed control algorithm forMH-FLto tune the D2D rounds in each cluster over time to meet specific convergence criteria. Our experiments on real-world datasets verify our analytical results and demonstrate the advantages ofMH-FLin terms of resource utilization metrics.

Device-to-device (D2D) communications is expected to be a critical enabler of distributed computing in edge networks at scale. A key challenge in providing this capability is the requirement for judicious management of the heterogeneous communication and computation resources that exist at the edge to meet processing needs. In this paper, we develop an optimization methodology that considers the network topology jointly with device and network resource allocation to minimize total D2D overhead, which we quantify in terms of time and energy required for task processing. Variables in our model include task assignment, CPU allocation, subchannel selection, and beamforming design for multiple-input multiple-output (MIMO) wireless devices. We propose two methods to solve the resulting non-convex mixed integer program: semi-exhaustive search optimization, which represents a “best-effort” at obtaining the optimal solution, and efficient alternate optimization, which is more computationally efficient. As a component of these two methods, we develop a novel coordinated beamforming algorithm which we show obtains the optimal beamformer for a common receiver characteristic. Through numerical experiments, we find that our methodology yields substantial improvements in network overhead compared with local computation and partially optimized methods, which validates our joint optimization approach. Further, we find that the efficient alternate optimization scales well with the number of nodes, and thus can be a practical solution for D2D computing in large networks.

Intelligent reflecting surfaces (IRS) consist of configurable meta-atoms, which can change the wireless propagation environment through design of their reflection coefficients. We consider a practical setting where (i) the IRS reflection coefficients are configured by adjusting tunable elements embedded in the meta-atoms, (ii) the IRS reflection coefficients are affected by the incident angles of the incoming signals, (iii) the IRS is deployed in multi-path, time-varying channels, and (iv) the feedback link from the base station to the IRS has a low data rate. Conventional optimization-based IRS control protocols, which rely on channel estimation and conveying the optimized variables to the IRS, are not applicable in this setting due to the difficulty of channel estimation and the low feedback rate. Therefore, we develop a novel adaptive codebook-based limited feedback protocol where only a codeword index is transferred to the IRS. We propose two solutions for adaptive codebook design, random adjacency (RA) and deep neural network policy-based IRS control (DPIC), both of which only require the end-to-end compound channels. We further develop several augmented schemes based on RA and DPIC. Numerical evaluations show that the data rate and average data rate over one coherence time are improved substantially by our schemes.

We study the problem of predicting student knowledge acquisition in online courses from clickstream behavior. Motivated by the proliferation of eLearning lecture delivery, we specifically focus on student in-video activity in lectures videos, which consist of content and in-video quizzes. Our methodology for predicting in-video quiz performance is based on three key ideas we develop. First, we model students’ clicking behavior via time-series learning architectures operating on raw event data, rather than defining hand-crafted features as in existing approaches that may lose important information embedded within the click sequences. Second, we develop a self-supervised clickstream pre-training to learn informative representations of clickstream events that can initialize the prediction model effectively. Third, we propose a clustering guided meta-learning-based training that optimizes the prediction model to exploit clusters of frequent patterns in student clickstream sequences. Through experiments on three real-world datasets, we demonstrate that our method obtains substantial improvements over two base-line models in predicting students’ in-video quiz performance. Further, we validate the importance of the pre-training and meta-learning components of our framework through ablation studies. Finally, we show how our methodology reveals insights on video-watching behavior associated with knowledge acquisition for useful learning analytics.

This research WIP contributes to understanding the relationship between learning engagement in Learning Management System (LMS) and outcomes in an online course. In large engineering courses, it is challenging for instructors to identify who is engaging with course materials at a level necessary to be successful in terms of course outcomes. The purpose of this research WIP study is two-fold: (1) to develop metrics for quantifying learner engagement in online courses, and (2) to explore the relationship between engagement and student success. Our research question is: How does learning engagement relate to course outcomes? We modeled learner engagement on a course level using the following features: number of views per content object, total time spent in the platform, percentage of the course accessed by the learners, percentage of feedback read, and number of attempts per quiz. We used the data collected by the LMS in a large first-year engineering course. We obtained data in Fall 2020, the first semester that many traditional universities were forced mostly or entirely online. After calculating the proposed metrics, we used a linear mixed model to analyze the effect of engagement on learning outcomes. Our linear mixed model shows that all engagement metrics are positively related to the final grade. However, the results also indicate that the relationship between engagement and learning outcomes is not linear; more complex modeling is needed to further explore this relationship.

Federated learning has emerged as a popular technique for distributing model training across the network edge. Its learning architecture is conventionally a star topology be-tween the devices and a central server. In this paper, we propose two timescale hybrid federated learning (TT-Hf),which migrates to a more distributed topology via device-to-device (D2D) communications. In TT-HF, local model training occurs at devices via successive gradient iterations, and the synchronization process occurs at two timescales: (i) macro-scale, where global aggregations are carried out via device-server interactions, and (ii) micro-scale, where local aggregations are carried out via D2D cooperative consensus formation in different device clusters. Our theoretical analysis reveals how device, cluster, and network-level parameters affect the convergence of TT-HF, and leads to a set of conditions under which a convergence rate of O(1/t) is guaranteed. Experimental results demonstrate the improvements in convergence and utilization that can be obtained by TT-HF over state-of-the-art federated learning baselines.

In general, reliable communication via multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) requires accurate channel estimation at the receiver. The existing literature largely focuses on denoising methods for channel estimation that depend on either (i) channel analysis in the time-domain with prior channel knowledge or (ii) supervised learning techniques which require large prelabeled datasets for training. To address these limitations, we present a frequency-domain denoising method based on a reinforcement learning framework that does not need a priori channel knowledge and pre-labeled data. Our methodology includes a new successive channel denoising process based on channel curvature computation, for which we obtain a channel curvature magnitude threshold to identify unreliable channel estimates. Based on this process, we formulate the denoising mechanism as a Markov decision process, where we define the actions through a geometry-based channel estimation update, and the reward function based on a policy that reduces mean squared error (MSE). We then resort to Q-learning to update the channel estimates. Numerical results verify that our denoising algorithm can successfully mitigate noise in channel estimates. In particular, our algorithm provides a significant improvement over the practical least squares (LS) estimation method and provides performance that approaches that of the ideal linear minimum mean square error (LMMSE) estimation with perfect knowledge of channel statistics.

Automatic modulation classification (AMC) aims to improve the efficiency of crowded radio spectrums by automatically predicting the modulation constellation of wireless RF signals. Recent work has demonstrated the ability of deep learning to achieve robust AMC performance using raw in-phase and quadrature (IQ) time samples. Yet, deep learning models are highly susceptible to adversarial interference, which cause intelligent prediction models to misclassify received samples with high confidence. Furthermore, adversarial interference is often transferable, allowing an adversary to attack multiple deep learning models with a single perturbation crafted for a particular classification network. In this work, we present a novel receiver architecture consisting of deep learning models capable of withstanding transferable adversarial interference. Specifically, we show that adversarial attacks crafted to fool models trained on time-domain features are not easily transferable to models trained using frequency-domain features. In this capacity, we demonstrate classification performance improvements greater than 30% on recurrent neural networks (RNNs) and greater than 50% on convolutional neural networks (CNNs). We further demonstrate our frequency feature-based classification models to achieve accuracies greater than 99% in the absence of attacks.

The conventional federated learning (FedL) architecture distributes machine learning (ML) across worker devices by having them train local models that are periodically aggregated by a server. FedL ignores two important characteristics of contemporary wireless networks, however: (i) the network may contain heterogeneous communication/computation resources, while (ii) there may be significant overlaps in devices' local data distributions. In this work, we develop a novel optimization methodology that jointly accounts for these factors via intelligent device sampling complemented by device-to-device (D2D) offloading. Our optimization aims to select the best combination of sampled nodes and data offloading configuration to maximize FedL training accuracy subject to realistic constraints on the network topology and device capabilities. Theoretical analysis of the D2D offloading subproblem leads to new FedL convergence bounds and an efficient sequential convex optimizer. Using this result, we develop a sampling methodology based on graph convolutional networks (GCNs) which learns the relationship between network attributes, sampled nodes, and resulting offloading that maximizes FedL accuracy. Through evaluation on real-world datasets and network measurements from our IoT testbed, we find that our methodology while sampling less than 5% of all devices outperforms conventional FedL substantially both in terms of trained model accuracy and required resource utilization.

Federated learning has emerged as a popular technique for distributing machine learning (ML) model training across the wireless edge. In this paper, we proposetwo timescale hybrid federated learning(TT-HF), a semi-decentralized learning architecture that combines the conventional device-to-server communication paradigm for federated learning with device-to-device (D2D) communications for model training. InTT-HF, during each global aggregation interval, devices (i) perform multiple stochastic gradient descent iterations on their individual datasets, and (ii) aperiodically engage in consensus procedure of their model parameters through cooperative, distributed D2D communications within local clusters. With a new general definition of gradient diversity, we formally study the convergence behavior ofTT-HF, resulting in new convergence bounds for distributed ML. We leverage our convergence bounds to develop an adaptive control algorithm that tunes the step size, D2D communication rounds, and global aggregation period ofTT-HFover time to target a sublinear convergence rate of$\mathcal {O}(1/t)$while minimizing network resource utilization. Our subsequent experiments demonstrate thatTT-HFsignificantly outperforms the current art in federated learning in terms of model accuracy and/or network energy consumption in different scenarios where local device datasets exhibit statistical heterogeneity. Finally, our numerical evaluations demonstrate robustness against outages caused by fading channels, as well favorable performance with non-convex loss functions.

Federated learning has emerged recently as a promising solution for distributing machine learning tasks through modern networks of mobile devices. Recent studies have obtained lower bounds on the expected decrease in model loss that is achieved through each round of federated learning. However, convergence generally requires a large number of communication rounds, which induces delay in model training and is costly in terms of network resources. In this paper, we propose a fast-convergent federated learning algorithm, called$\mathsf {FOLB}$, which performs intelligent sampling of devices in each round of model training to optimize the expected convergence speed. We first theoretically characterize a lower bound on improvement that can be obtained in each round if devices are selected according to the expected improvement their local models will provide to the current global model. Then, we show that$\mathsf {FOLB}$obtains this bound through uniform sampling by weighting device updates according to their gradient information.$\mathsf {FOLB}$is able to handle both communication and computation heterogeneity of devices by adapting the aggregations according to estimates of device’s capabilities of contributing to the updates. We evaluate$\mathsf {FOLB}$in comparison with existing federated learning algorithms and experimentally show its improvement in trained model accuracy, convergence speed, and/or model stability across various machine learning tasks and datasets.

Existing topic modeling and text segmentation methodologies generally require large datasets for training, limiting their capabilities when only small collections of text are available. In this work, we reexamine the inter-related problems of “topic identification” and “text segmentation” for sparse document learning, when there is a single new text of interest. In developing a methodology to handle single documents, we face two major challenges. First is sparse information : with access to only one document, we cannot train traditional topic models or deep learning algorithms. Second is significant noise : a considerable portion of words in any single document will produce only noise and not help discern topics or segments. To tackle these issues, we design an unsupervised, computationally efficient methodology called Biclustering Approach to Topic modeling and Segmentation (BATS). BATS leverages three key ideas to simultaneously identify topics and segment text: (i) a new mechanism that uses word order information to reduce sample complexity, (ii) a statistically sound graph-based biclustering technique that identifies latent structures of words and sentences, and (iii) a collection of effective heuristics that remove noise words and award important words to further improve performance. Experiments on six datasets show that our approach outperforms several state-of-the-art baselines when considering topic coherence, topic diversity, segmentation, and runtime comparison metrics.

Fog computing promises to enable machine learning tasks to scale to large amounts of data by distributing processing across connected devices. Two key challenges to achieving this goal are (i) heterogeneity in devices’ compute resources and (ii) topology constraints on which devices communicate with each other. We address these challenges by developing a novel network-aware distributed learning methodology where devices optimally share local data processing and send their learnt parameters to a server for periodic aggregation. Unlike traditional federated learning, our method enables devices to offload their data processing tasks to each other, with these decisions optimized to trade off costs associated with data processing, offloading, and discarding. We analytically characterize the optimal data transfer solution under different assumptions on the fog network scenario, showing for example that the value of offloading is approximately linear in the range of computing costs in the network when the cost of discarding is modeled as decreasing linearly in the amount of data processed at each node. Our experiments on real-world data traces from our testbed confirm that our algorithms improve network resource utilization substantially without sacrificing the accuracy of the learned model, for varying distributions of data across devices. We also investigate the effect of network dynamics on model learning and resource costs.

Federated learning has received significant attention as a potential solution for distributing machine learning (ML) model training through edge networks. This work addresses an important consideration of federated learning at the network edge: communication delays between the edge nodes and the aggregator. A technique called FedDelAvg (federated delayed averaging) is developed, which generalizes the standard federated averaging algorithm to incorporate a weighting between the current local model and the delayed global model received at each device during the synchronization step. Through theoretical analysis, an upper bound is derived on the global model loss achieved by FedDelAvg, which reveals a strong dependency of learning performance on the values of the weighting and learning rate. Experimental results on a popular ML task indicate significant improvements in terms of convergence speed when optimizing the weighting scheme to account for delays.

In this paper, we study the distributed computational capabilities of device-to-device (D2D) networks. A key characteristic of D2D networks is that their topologies are reconfigurable to cope with network demands. For distributed computing, resource management is challenging due to limited network and communication resources, leading to inter-channel interference. To overcome this, recent research has addressed the problems of wireless scheduling, subchannel allocation, power allocation, and multiple-input multiple-output (MIMO) signal design, but has not considered them jointly. In this paper, unlike previous mobile edge computing (MEC) approaches, we propose a joint optimization of wireless MIMO signal design and network resource allocation to maximize energy efficiency. Given that the resulting problem is a non-convex mixed integer program (MIP) which is prohibitive to solve at scale, we decompose its solution into two parts: (i) a resource allocation subproblem, which optimizes the link selection and subchannel allocations, and (ii) MIMO signal design subproblem, which optimizes the transmit beamformer, transmit power, and receive combiner. Simulation results using wireless edge topologies show that our method yields substantial improvements in energy efficiency compared with cases of no offloading and partially optimized methods and that the efficiency scales well with the size of the network.

Fog computing promises to enable machine learning tasks to scale to large amounts of data by distributing processing across connected devices. Two key challenges to achieving this are (i) heterogeneity in devices' compute resources and (ii) topology constraints on which devices can communicate. We are the first to address these challenges by developing a network-aware distributed learning optimization methodology where devices process data for a task locally and send their learnt parameters to a server for aggregation at certain time intervals. Unlike traditional federated learning frameworks, our method enables devices to offload their data processing tasks, with these decisions determined through a convex data transfer optimization problem that trades off costs associated with devices processing, offloading, and discarding data points. We analytically characterize the optimal data transfer solution for different fog network topologies, showing for example that the value of a device offloading is approximately linear in the range of computing costs in the network. Our subsequent experiments on both synthetic and real-world datasets we collect confirm that our algorithms are able to improve network resource utilization substantially without sacrificing the accuracy of the learned model.

Growing mobile data usage has led to end users paying substantial data costs, while Internet service providers (ISPs) struggle to upgrade their networks to keep up with demand and maintain high quality-of-service (QoS). This problem is particularly severe for smaller ISPs with less capital. Instead of simply upgrading their network infrastructure, ISPs can pool their networks to provide a good QoS and attract more users. Such a vISP (virtual ISP), for example, Google's Project Fi, allows users to access any of its partner ISPs' networks. We provide the first systematic analysis of a vISP's economic impact, showing that the vISP provides a viable solution for smaller ISPs attempting to attract more users, but may not maintain a positive profit if users' data demands evolve. To do so, we consider users' decisions of whether to defect from their current ISP to the vISP, as well as existing ISPs' decisions on whether to partner with the vISP. We derive the vISP's dependence on user behavior and partner ISPs: users with very light or very heavy usage are the most likely to defect, while ISPs with heavy-usage customers can benefit from declining to partner with the vISP. Our analytical results are verified with extensive numerical simulations.

We consider the problem of jointly predicting the quality and timing of responses to questions asked in online discussion forums. While prior work has focused on identifying users most likely to answer and/or to provide the highest quality answers to a question, the promptness of the response is also a key factor of user satisfaction. To address this, we propose point process and neural network-based algorithms for three prediction tasks regarding a user's response to a question: whether the user will answer, the net votes that will be received on the answer, and the time that will elapse before the answer. These algorithms learn over a set of 20 features we define for each pair of user and question that quantify both topical and structural aspects of the forums, including discussion post similarities and social centrality measures. Through evaluation on a Stack Overflow dataset consisting of 20,000 question threads, we find that our method outperforms baselines on each prediction task by more than 20%. We also find that the importance of the features varies depending on the task and the amount of historical data available for inference. At the end, we design a question recommendation system that incorporates these predictions to jointly optimize response quality and timing in forums subject to user constraints.

It is of crucial importance to simultaneously protect against sensitive attributes in data while building predictive models. In this paper, we tackle the problem of learning representations from raw data that are i) informative and predictive of desirable variables, and ii) private and protect against adversaries that attempt to recover sensitive variables. We cast this problem under the generative adversarial network (GAN) framework and design three components: an encoder, an ally that predicts the desired variables, and an adversary that predicts the sensitive ones. As a use case, we apply our approach to learn representations of raw student clickstream event data captured as they watch lecture videos in massive open online courses (MOOCs). Through experiments on a real-world dataset collected from a MOOC, we demonstrate that our method can learn a low-dimensional representation of each user that i) excels at classifying whether a user will answer a quiz question correctly, and ii) prevents an adversary from recovering each user's identity. Our results indicate that our approach is effective in learning representations that are both informative and private.

We consider the problem of predicting link formation in Social Learning Networks (SLN), a type of social network that forms when people learn from one another through structured interactions. While link prediction has been studied for general types of social networks, the evolution of SLNs over their lifetimes coupled with their dependence on which topics are being discussed presents new challenges for this type of network. To address these challenges, we develop a time-series prediction methodology that uses a recurrent neural network architecture to pass network state between time periods, and that models over three types of SLN features updated in each period: neighborhood-based (e.g., resource allocation), path-based (e.g., shortest path), and post-based (e.g., topic similarity). Through evaluation on four real-world datasets from Massive Open Online Course (MOOC) discussion forums, we find that our method obtains substantial improvements over a Bayesian model and an unsupervised baseline, with AUCs typically above 0.75 and reaching 0.97 depending on the dataset. Our feature importance analysis shows that while neighborhood-based features contribute the most to the results, post-based and path-based features add additional information that significantly improve the predictions. We also find that several input features have opposite directions of correlation between link formation and post quality, suggesting that response time and quality are two competing objectives to be accounted for in SLN link recommendation systems.

We study learning outcome prediction for online courses. Whereas prior work has focused on semester-long courses with frequent student assessments, we focus on short-courses that have single outcomes assigned by instructors at the end. The lack of performance data makes the behavior of learners, captured as they interact with course content and with one another in Social Learning Networks (SLN), essential for prediction. Our method defines several (machine) learning features based on behaviors collected on the modes of (human) learning in a course, and uses them in appropriate classifiers. Through evaluation on data captured from three two-week courses hosted through our delivery platforms, we make three key observations: (i) behavioral data is predictive of learning outcomes in short-courses (our classifiers achieving AUCs ≥ 0.8 after the two weeks), (ii) it has an early detection capability (AUCs ≥ 0.7 with the first week of data), and (iii) the content features have an “earliest” detection capability (with higher AUC in the first few days), while the SLN features become the more predictive set over time, as the network matures. We also discuss how our method can generate behavioral analytics for instructors.

Growing mobile data usage has led to end users paying substantial data costs, while Internet service providers (ISPs) struggle to upgrade their networks to keep up with demand and maintain high quality-of-service (QoS). This problem is particularly severe for smaller ISPs with less capital. Instead of simply upgrading their network infrastructure, ISPs can pool their networks to provide a good QoS and attract more users. Such a vISP (virtual ISP), for example, Google's Project Fi, allows users to access any of its partner ISPs' networks. We provide the first systematic analysis of a vISP's economic impact, showing that the vISP provides a viable solution for smaller ISPs attempting to attract more users, but may not maintain a positive profit if users' data demands evolve. To do so, we consider users' decisions of whether to defect from their current ISP to the vISP, as well as ISPs' decisions on whether to partner with the vISP. We derive the vISP's dependence on user behavior and partner ISPs: users with very light or very heavy usage are the most likely to defect, while ISPs with heavy-usage customers can benefit from declining to partner with the vISP. Our analytical results are verified with extensive numerical simulations.

A Social Learning Network (SLN) emerges when users exchange information on educational topics with structured interactions. The recent proliferation of massively scaled online (human) learning, such as Massive Open Online Courses (MOOCs), has presented a plethora of research challenges surrounding SLN. In this paper, we ask: How efficient are these networks? We propose a framework in which SLN efficiency is determined by comparing user benefit in the observed network to a benchmark of maximum utility achievable through optimization. Our framework defines the optimal SLN through utility maximization subject to a set of constraints that can be inferred from the network. Through evaluation on four MOOC discussion forum datasets and optimizing over millions of variables, we find that SLN efficiency can be rather low (from 68% to 82% depending on the specific parameters and dataset), which indicates that much can be gained through optimization. We find that the gains in global utility (i.e., average across users) can be obtained without making the distribution of local utilities (i.e., utility of individual users) less fair. We also discuss ways of realizing the optimal network in practice, through curated news feeds in online SLN.

Cloud service providers (CSPs) often face highly dynamic user demands for their resources, which can make it difficult for them to maintain consistent quality-of-service. Some CSPs try to stabilize user demands by offering sustained-use discounts to jobs that consume more instance-hours per month. These discounts present an opportunity for users to pool their usage together into a single ``job.'' In this paper, we examine the viability of a middleman, the cloud virtual service provider (CVSP), that rents cloud resources from a CSP and then resells them to users. We show that the CVSP's business model is only viable if the average job runtimes and thresholds for sustained-use discounts are sufficiently small; otherwise, the CVSP cannot simultaneously maintain low job waiting times while qualifying for a sustained-use discount. We quantify these viability conditions by modeling the CVSP's job scheduling and then use this model to derive users' utility-maximizing demands and the CVSP's profit-maximizing price, as well as the optimal number of instances that the CVSP should rent from the CSP. We verify our results on a one-month trace from Google's production compute cluster, through which we first validate our assumptions on the job arrival and runtime distributions, and then show that the CVSP is viable under these workload traces. Indeed, the CVSP can earn a positive profit without significantly impacting the CSP's revenue, indicating that the CSP and CVSP can coexist in the cloud market.

We study student performance prediction in Massive Open Online Courses (MOOCs), where the objective is to predict whether a user will be Correct on First Attempt (CFA) in answering a question. In doing so, we develop novel techniques that leverage behavioral data collected by MOOC platforms. Using video-watching clickstream data from one of our MOOCs, we first extract summary quantities (e.g., fraction played, number of pauses) for each user-video pair, and show how certain intervals/sets of values for these behaviors quantify that a pair is more likely to be CFA or not for the corresponding question. Motivated by these findings, our methods are designed to determine suitable intervals from training data and to use the corresponding success estimates as learning features in prediction algorithms. Tested against a large set of empirical data, we find that our schemes outperform standard algorithms (i.e., without behavioral data) for all datasets and metrics tested. Moreover, the improvement is particularly pronounced when considering the first few course weeks, demonstrating the “early detection” capability of such clickstream data. We also discuss how CFA prediction can be used to depict graphs of the Social Learning Network (SLN) of students, which can help instructors manage courses more effectively.