Temporal knowledge graph question answering (TKGQA) involves multi-hop reasoning over temporally constrained entity relationships in the knowledge graph to answer a given question. However, at each hop, large language models (LLMs) retrieve subgraphs with numerous temporally similar and semantically complex relations, increasing the risk of suboptimal decisions and error propagation. To address these challenges, we propose the multi-hop reasoning enhanced (MRE) framework, which enhances both forward and backward reasoning to improve the identification of globally optimal reasoning trajectories. Specifically, MRE begins with prompt engineering to guide LLM in generating diverse reasoning trajectories for the given question. Valid reasoning trajectories are then selected for supervised fine-tuning, serving as a cold-start strategy. Finally, we introduce Tree-Group Relative Policy Optimization (T-GRPO)—a recursive, tree-structured learning-by-exploration approach. At each hop, exploration establishes strong causal dependencies on the previous hop, while evaluation is informed by multi-path exploration feedback from subsequent hops. Experimental results on two TKGQA benchmarks indicate that the proposed MRE-based model consistently surpasses state-of-the-art (SOTA) approaches in handling complex multi-hop queries. Further analysis highlights improved interpretability and robustness to noisy temporal annotations.

A Wafer-scale GPU connects a large number of chiplets via a high-bandwidth, low-latency interposer-based network, promising to overcome the communication bottleneck of traditional multi-GPU systems. While prior work has prototyped wafer-scale GPUs to demonstrate technical feasibility, scaling to massive chiplet counts creates new bottlenecks: virtual-tophysical address translation becomes severely constrained by massive concurrent requests and long multi-hop network latencies. We propose HDPAT, a hardware-accelerated distributed address translation system that addresses this challenge through three complementary techniques: (1) Concentric caching converts near-IOMMU chiplets into hierarchical translation caches based on their distance to the IOMMU. A lightweight rotation mechanism ensures that there is always a nearby chiplet that can provide translation caching. (2) The redirection table further reduces the burden of IOMMU by delegating translations to caching chiplets, and (3) Prefetching proactively delivers potentially needed address translation into the chiplet to improve translation cache hit rate. Experimental results on 14 representative workloads show that HDPAT improves overall performance by an average of$1.57 \times$.

As process nodes shrink, intensified IR drop and electromigration (EM) issues, driven by the power delivery network (PDN), affect reliability and further compress the routability of the entire design. Since physical design is time-consuming, effectively predicting and optimizing PDN during the power planning stage is crucial. This effort is complicated by complex interactions between multiple parameters and the challenges associated with nonconvex optimization. To overcome these limitations, we propose a learning-driven hierarchical particle swarm optimization framework for PDN, integrated with a residual TransUNet-based prediction model for post-placement congestion, EM, and IR drop prediction. In addition, we introduce the gate network for loss fusion, which excels in balancing output predictions and enhancing the accuracy of critical tasks. Analogous to hyperparameter optimization in deep learning, we can utilize GPU parallelization to search for the global optimum of the PDN structure efficiently. In circuits at the 22nmand 16nmtechnology nodes, with cell counts ranging from 560 to nearly 200, 000, our method achieves reductions in mean and maximum congestion by 15% and 49%, respectively, compared to state-of-the-art methods. In addition, it achieves an overall optimization time of around 22 seconds.

As semiconductor processes advance, the power delivery network (PDN) increasingly affects the power supply from the pads to the cells. Significant IR drops in standard cells can lead to timing violations, while suboptimal PDN topologies can lead to increased congestion. Together, these factors degrade overall chip performance and reliability. To accelerate design iteration, accurately and efficiently predicting unevenly distributed IR drop and congestion, especially in hotspot areas, has become a critical challenge. This article introduces an enhanced TransUNet-based framework for distribution-aware static IR drop and congestion prediction, treating both problems as separate but related image prediction tasks. Such an abstraction preserves the IR drop and congestion distribution patterns on the original real physical layout and retains local hot spots. Our proposed framework leverages image classification techniques to model IR drop prediction as a spatial pattern recognition task, effectively addressing the long-tail distribution in different regions. To enhance hotspot prediction, we incorporate wavelet transform and transformer-based analysis to enable multiscale feature fusion. In the open-source CircuitNet dataset, our method predicts static IR drop with a mean absolute error( MAE ) of 0.374 mV and a maximum error rate( Err m ) of 18.7%, reducing MAE and Err m by 77.4% and 79.2%, respectively, compared to the state-of-the-art method, all within 100 ms. The congestion prediction evaluations show 65.3% lower NRMSE scores and 18.4% higher SSIM scores relative to the existing SOTA approach. Our approach accurately and reliably predicts long-tail distributions and localized hotspots in both IR drop and congestion tasks.

The accurate prediction of genedisease associations is essential for understanding pathogenic mechanisms and identifying therapeutic targets. While existing computational methods have integrated diverse biological information, they struggle with the sparsity of known associations and the lack of reliable negative samples. To address these issues, we propose NCVGAE, a novel framework that extends the Variational Graph Auto-Encoder (VGAE) by incorporating a hybrid negative sampling strategy and contrastive learning to enhance representation learning under sparse supervision. We devise a Hybrid Negative Sampling strategy that combines similarity-guided selection-using topological and biological similarity to identify high-confidence negatives-with random sampling to ensure diversity. This helps reduce noise from unreliable negatives and improves the quality of supervision. The encoder module employs a Graph Neural Network (GNN) to effectively learn robust embeddings for genes and diseases. During the training phase, a contrastive learning objective refines embeddings by pulling together true associations and pushing apart negatives. A multilayer perceptron then predicts association scores. Experiments on three public datasets show that NCVGAE consistently outperforms state-of-the-art methods in terms of AUC and AUPR, highlighting its effectiveness and generalizability.

Emerging from the pairwise attention in conventional Transformers, there is a growing interest in sparse attention mechanisms that align more closely with localized, contextual learning in the biological brain. Existing studies such as the Coordination method employ iterative cross-attention mechanisms with a bottleneck to enable the sparse association of inputs. However, these methods are parameter inefficient and fail in more complex relational reasoning tasks. To this end, we propose Associative Transformer (AiT) to enhance the association among sparsely attended input tokens, improving parameter efficiency and performance in various vision tasks such as classification and relational reasoning. AiT leverages a learnable explicit memory comprising specialized priors that guide bottleneck attentions to facilitate the extraction of diverse localized tokens. Moreover, AiT employs an associative memory-based token reconstruction using a Hopfield energy function. The extensive empirical experiments demonstrate that AiT requires significantly fewer parameters and attention layers outperforming a broad range of sparse Transformer models. Additionally, AiT outperforms the SOTA sparse Transformer models including the Coordination method on the Sort-of-CLEVR dataset.

Conventional neural networks including Transformers encounter the catastrophic forgetting problem during sequential task learning, where learning new tasks interferes with previously learned knowledge. Existing memory replay and regularization methods cannot effectively eliminate interference among different tasks. Soft parameter sharing methods usually necessitate a large amount of additional parameters for learning each task, and task identity information is essential for leveraging task-specific parameters. To this end, we propose Remembering Transformer leveraging an adapter mixtures architecture enhanced by a generative routing mechanism for efficient task retention. In generative routing, input samples are allocated to the most relevant expert adapters based on a reconstruction loss. Moreover, unlike existing studies on soft parameter sharing that do not consider model capacity limitations, we investigate a challenging setting where the number of task-specific parameters is constrained. In particular, we devise an adapter fusion strategy to aggregate resembling experts based on similarity matching and knowledge distillation. Extensive empirical results measured by task accuracy, forgetting rate, and memory footprint, demonstrate that Remembering Transformer significantly enhances knowledge retention without task identity information. The proposed method surpasses various conventional methods with enhanced parameter efficiency in a broad range of incremental learning tasks.

Dynamic Binary instrumentation (DBI) is a widely used technique for collecting detailed, fine-grained information from program execution without requiring recompilation or access to the program's source code. DBI provides several benefits over static instrumentation, including full code discovery and the ability to selectively toggle profiling during runtime. Luthier is an open-source DBI framework targeting AMD GPUs, designed to integrate and run seamlessly on the ROCm software stack. During runtime, Luthier allows inspection of loaded GPU code objects and carries out instrumentation by either manually modifying instructions or inserting calls to special device functions (i.e., 'hooks'') at user-specified locations in the program. Luthier hooks allow inspection and modification of the device visible state, and can communicate with the host via host-accessible device memory buffers. Luthier also supports switching between instrumented and un-instrumented versions of a kernel. In this paper, we describe some of the key design challenges we encountered when developing this open-source DBI framework. We then showcase Luthier's user-facing APIs and internal components, providing example usecases implemented using our framework. While Luthier incurs a 50X runtime overhead (on average) when running an instrumented application, this overhead is 10 times lower as compared to the state-of-the-art GPU-based DBI framework, when running equivalent tools on the same workload written in CUDA.

Visual Question Answering (VQA) based on multi-modal data facilitates real-life applications such as home robots and medical diagnoses. One significant challenge is to devise a robust decentralized learning framework for various client models where centralized data collection is refrained due to confidentiality concerns. This work aims to tackle privacy-preserving VQA by decoupling a multi-modal model into representation modules and a contrastive module, leveraging inter-module gradients sharing and inter-client weight sharing. To this end, we propose Bidirectional Contrastive Split Learning (BiCSL) to train a global multi-modal model on the entire data distribution of decentralized clients. We employ the contrastive loss that enables a more efficient self-supervised learning of decentralized modules. Comprehensive experiments are conducted on the VQA-v2 dataset based on five SOTA VQA models, demonstrating the effectiveness of the proposed method. Furthermore, we inspect BiCSL's robustness against a dual-key backdoor attack on VQA. Consequently, BiCSL shows significantly enhanced resilience when exposed to the multi-modal adversarial attack compared to the centralized learning method, which provides a promising approach to decentralized multi-modal learning.

Trojan attacks embed perturbations in input data leading to malicious behavior in neural network models. A combination of various Trojans in different modalities enables an adversary to mount a sophisticated attack on multimodal learning such as Visual Question Answering (VQA). However, multimodal Trojans in conventional methods are susceptible to parameter adjustment during processes such as fine-tuning. To this end, we propose an instance-level multimodal Trojan attack on VQA that efficiently adapts to fine-tuned models through a dual-modality adversarial learning method. This method compromises two specific neurons in a specific perturbation layer in the pretrained model to produce overly large neuron activations. Then, a malicious correlation between these overactive neurons and the malicious output of a fine-tuned model is established through adversarial learning. Extensive experiments are conducted using the VQA-v2 dataset, based on a wide range of metrics including sample efficiency, stealthiness, and robustness. The proposed attack demonstrates enhanced performance with diverse vision and text Trojans tailored for each sample. We demonstrate that the proposed attack can be efficiently adapted to different fine-tuned models, by injecting only a few shots of Trojan samples. Moreover, we investigate the attack performance under conventional defenses, where the defenses cannot effectively mitigate the attack.

Meta-learning usually refers to a learning algorithm that learns from other learning algorithms. The problem of uncertainty in the predictions of neural networks shows that the world is only partially predictable and a learned neural network cannot generalize to its ever-changing surrounding environments. Therefore, the question is how a predictive model can represent multiple predictions simultaneously. We aim to provide a fundamental understanding of learning to learn in the contents of Decentralized Neural Networks (Decentralized NNs) and we believe this is one of the most important questions and prerequisites to building an autonomous intelligence machine. To this end, we shall demonstrate several pieces of evidence for tackling the problems above with Meta Learning in Decentralized NNs. In particular, we will present three different approaches to building such a decentralized learning system: (1) learning from many replica neural networks, (2) building the hierarchy of neural networks for different functions, and (3) leveraging different modality experts to learn cross-modal representations.

Anomaly detection is an important function in IoT applications for finding outliers caused by abnormal events. Anomaly detection sometimes comes with high-frequency data sampling which should be carried out at Edge devices rather than Cloud. In this paper, we consider the case that multiple IoT devices are installed in a single remote site and that they collaboratively detect anomalies from the observations with device-to-device communications. For this, we propose a fully distributed collaborative scheme for training distributed anomaly detectors. We introduce the concept of Global Anomaly which sample is not only rare to the local device but rare to all the devices in the target domain. We also propose a distributed threshold-finding algorithm for Global Anomaly detection. With our standard benchmark-based evaluation, we have confirmed that our scheme trained anomaly detectors perfectly across the devices. We have also confirmed that the devices collaboratively found thresholds for Global Anomaly detection with low false positive rates while achieving high true positive rates with few exceptions.

The rapidly increasing number of smart devices on the Internet necessitates an efficient inspection system for safeguarding our networks from suspicious activities such as Address Resolution Protocol (ARP) probes. In this research, we analyze sequence data of ARP traffic on LAN based on the numerical count and degree of its packets. A dynamic threshold is employed to detect underlying suspicious activities, which are further converted into feature vectors to train an unsupervised autoencoder neural network. Then, we leverage K-means clustering to separate the extracted latent features of suspicious activities from the autoencoder into various patterns.

Model poisoning attacks on federated learning intrude in the entire system via compromising an edge model, resulting in malfunctioning of machine learning models. Such compromised models are tampered with to perform adversary-desired behaviors. In particular, we considered a semi-targeted situation where the source class is predetermined however the target class is not. The goal is to cause the global classifier to misclassify data of the source class. Though approaches such as label flipping have been adopted to inject poisoned parameters into federated learning, it has been shown that their performances are usually class-sensitive varying with different target classes applied. Typically, an attack can become less effective when shifting to a different target class. To overcome this challenge, we propose the Attacking Distance-aware Attack (ADA) to enhance a poisoning attack by finding the optimized target class in the feature space. Moreover, we studied a more challenging situation where an adversary had limited prior knowledge about a client's data. To tackle this problem, ADA deduces pair-wise distances between different classes in the latent feature space from shared model parameters based on the backward error analysis. We performed extensive empirical evaluations on ADA by varying the factor of attacking frequency in three different image classification tasks. As a result, ADA succeeded in increasing the attack performance by 1.8 times in the most challenging case with an attacking frequency of 0.01.

With increasingly more sophisticated phishing campaigns in recent years, phishing emails lure people using more legitimate-looking personal contexts. To tackle this problem, instead of traditional heuristics-based algorithms, more adaptive detection systems such as natural language processing (NLP)powered approaches are essential to understanding phishing text representations. Nevertheless, concerns surrounding the collection of phishing data that might cover confidential information hinder the effectiveness of model learning. We propose a decentralized phishing email detection framework called Federated Phish Bowl (FedPB) which facilitates collaborative phishing detection with privacy. In particular, we devise a knowledge-sharing mechanism with federated learning (FL). Using long short-term memory (LSTM) for phishing detection, the framework adapts by sharing a global word embedding matrix across the clients, with each client running its local model with Non-IID data. We collected the most recent phishing samples to study the effectiveness of the proposed method using different client numbers and data distributions. The results show that FedPB can attain a competitive performance with a centralized phishing detector, with generality to various cases of FL retaining a prediction accuracy of 83%.

In this article, we presented a visualization method for representing network traffic features using raw data of it. The raw network traffic data was divided into regulated segments. By employing a supervised neural network and an expert-knowledge based labeling method, model training was conducted based on a dataset covering two weeks' network traffic, where the first week's data was employed as the training set and the second week's data was used as the validation set. At last, we achieved validation precision scores of 0.980 for detecting the ARP flooding, 0.800 and 0.815 for detecting the malicious SMB and TCP SYN flooding respectively.

With society's increasing dependence on the Internet, more private data is transmitted through networks every day. Unfortunately, this traffic is susceptible to a wide range of threats and vulnerabilities, including phishing attacks that trick users into compromising their systems or revealing sensitive personal information. In this research, we proposed a deep learning approach to detect malware using data collected from a web crawler that systematically sent requests to benign and malicious websites on the Internet. After applying procedures to segment the network flows and extract features, we used these extracted high-level network traffic features to train a deep neural network to recognize benign and malicious flows. Finally, we evaluated our malware detection approach against various metrics, including precision, recall, and f1 score. The achieved f1 score was 0.924, validating the overall performance of the detection scheme.

An attack on deep learning systems where intelligent machines collaborate to solve problems could cause a node in the network to make a mistake on a critical judgment. At the same time, the security and privacy concerns of AI have galvanized the attention of experts from multiple disciplines. In this research, we successfully mounted adversarial attacks on a federated learning (FL) environment using three different datasets. The attacks leveraged generative adversarial networks (GANs) to affect the learning process and strive to reconstruct the private data of users by learning hidden features from shared local model parameters. The attack was target-oriented drawing data with distinct class distribution from the CIFAR-10, MNIST, and Fashion-MNIST respectively. Moreover, by measuring the Euclidean distance between the real data and the reconstructed adversarial samples, we evaluated the performance of the adversary in the learning processes in various scenarios. At last, we successfully reconstructed the real data of the victim from the shared global model parameters with all the applied datasets.

With the approach of 5G Society, more and more devices have been connected to the Internet, where information is stored, analyzed, and shared. Federated learning allows participants to train a machine learning model through sharing the parameters of it based on local training, instead of raw private data at local. In this research, we propose the implementation of the blockchain in federated learning for local parameters evaluation and global parameter aggregation, thus alleviating the influence of end-point adversarial training data. Besides, all updates of local parameters are encrypted and stored in a block of the blockchain after the consensus by the committee. We evaluate the performance of the scheme when adopting various types of corruption to the adversary's dataset, including noise with various degrees and circle occlusion with various diameters. At last, it shows robust and resilient performance compared with the traditional federated learning, achieving a validation accuracy rate of 0.957 when adding noise with a degree of 1.0, and one of 0.944 when adopting circle occlusion with a diameter of 28 pixels for the classification.

Traditional approaches to cybersecurity issues usually protect users from attacks after the occurrence of specific types of attacks. Besides, patterns of recent cyberattacks tend to be changeable, which add up to unpredictability of them. On the other hand, machine learning, as a new method used to detect intrusion, is attracting more and more attention. Moreover, through the sharing of local training data, the centralized learning approach has proven to improve a model's performance. In this research, a segmented federated learning is proposed, different from a collaborative learning based on single global model in a traditional federated learning model, it keeps multiple global models which allow each segment of participants to conduct collaborative learning separately and rearranges the segmentation of participants dynamically as well. Furthermore, these multiple global models interact with each other for updating parameters, thus being adaptable to various participants' LANs. A dataset covering two months' traffic data from 20 participants' LANs in the LAN-Security Monitoring Project is used. We adopt three types of knowledge-based methods for labeling network events and train a CNN model based on the dataset. At last, we achieve validation accuracies of 0.923, 0.813 and 0.877 individually with these labeling methods.