In mobile edge computing (MEC) networks, mobile users generate diverse machine learning tasks dynamically over time. These tasks are typically offloaded to the nearest available edge server, by considering communication and computational efficiency. However, its operation does not ensure that each server specializes in a specific type of tasks and leads to severe overfitting or catastrophic forgetting of previous tasks. To improve the continual learning (CL) performance of online tasks, we are the first to introduce mixture-of-experts (MoE) theory in MEC networks and save MEC operation from the increasing generalization error over time. Our MoE theory treats each MEC server as an expert and dynamically adapts to changes in server availability by considering data transfer and computation time. Unlike existing MoE models designed for offline tasks, ours is designed to handle continuous streams of tasks in the MEC environment. We introduce an adaptive gating network in MEC to adaptively identify and route newly arrived tasks of unknown data distributions to available experts, enabling each expert to specialize in a specific type of task upon convergence. We first consider the MoE model with the simple but fundamental top-1 routing strategy to derive the minimum number of experts required to match each task with a specialized, available expert. Our MoE approach consistently reduces the overall generalization error over time, unlike the traditional MEC approach. Interestingly, when the number of experts is sufficient to ensure convergence, adding more experts delays the convergence time and worsens the generalization error. Furthermore, we extend our theoretical analysis to the more general MoE model with top-$k$routing to examine the effect of the number of selected experts$k$for each task. Our results demonstrate that, while top-$k$routing accelerates system convergence compared to the top-1 strategy, it may increase the overall generalization error due to the heavier workload on each expert. Finally, we perform extensive experiments on real datasets in deep neural networks (DNNs) to verify our theoretical results.

Continual learning (CL) has garnered significant attention because of its ability to adapt to new tasks that arrive over time. Catastrophic forgetting (of old tasks) has been identified as a major issue in CL, as the model adapts to new tasks. The Mixture-of-Experts (MoE) model has recently been shown to effectively mitigate catastrophic forgetting in CL, by employing a gating network to sparsify and distribute diverse tasks among multiple experts. However, there is a lack of theoretical analysis of MoE and its impact on the learning performance in CL. This paper provides the first theoretical results to characterize the impact of MoE in CL via the lens of overparameterized linear regression tasks. We establish the benefit of MoE over a single expert by proving that the MoE model can diversify its experts to specialize in different tasks, while its router learns to select the right expert for each task and balance the loads across all experts. Our study further suggests an intriguing fact that the MoE in CL needs to terminate the update of the gating network after sufficient training rounds to attain system convergence, which is not needed in the existing MoE studies that do not consider the continual task arrival. Furthermore, we provide explicit expressions for the expected forgetting and overall generalization error to characterize the benefit of MoE in the learning performance in CL. Interestingly, adding more experts requires additional rounds before convergence, which may not enhance the learning performance. Finally, we conduct experiments on both synthetic and real datasets to extend these insights from linear models to deep neural networks (DNNs), which also shed light on the practical algorithm design for MoE in CL.

The binary exponential backoff scheme is widely used in WiFi 7 and still incurs poor throughput performance under dynamic channel environments. Recent model-based approaches (e.g., non-persistent and p-persistent CSMA) simply optimize backoff strategies under a known and fixed node density, still leading to a large throughput loss due to inaccurate node density estimation. This paper is the first to propose LLM transformer-based in-context learning (ICL) theory for optimizing channel access. We design a transformer-based ICL optimizer to pre-collect collision-threshold data examples and a query collision case. They are constructed as a prompt as the input for the transformer to learn the pattern, which then generates a predicted contention window threshold (CWT). To train the transformer for effective ICL, we develop an efficient algorithm and guarantee a near-optimal CWT prediction within limited training steps. As it may be hard to gather perfect data examples for ICL in practice, we further extend to allow erroneous data input in the prompt. We prove that our optimizer maintains minimal prediction and throughput deviations from the optimal values. Experimental results on NS-3 further demonstrate our approach's advantage.

Today navigation applications (e.g., Waze and Google Maps) enable human users to learn and share the latest traffic observations, yet such information sharing simply aids selfish users to predict and choose the shortest paths to jam each other. Prior routing game studies focus on myopic users in oversimplified one-shot scenarios to regulate selfish routing via information hiding or pricing mechanisms. For practical human-in-the-loop learning (HILL) in repeated routing games, we face non-myopic users of differential past observations and need new mechanisms (preferably non-monetary) to persuade users to adhere to the optimal path recommendations. We model the repeated routing game in a typical parallel transportation network, which generally contains one deterministic path and$N$stochastic paths. We first prove that no matter under the information sharing mechanism in use or the latest routing literature’s hiding mechanism, the resultant price of anarchy (PoA) for measuring the efficiency loss from social optimum can approach infinity, telling arbitrarily poor exploration-exploitation tradeoff over time. Then we propose a novel user-differential probabilistic recommendation (UPR) mechanism to differentiate and randomize path recommendations for users with differential learning histories. We prove that our UPR mechanism ensures interim individual rationality for all users and significantly reduces$\text{PoA}=\infty$to close-to-optimal$\text{PoA}=1+\frac{1}{4N+3}$, which cannot be further reduced by any other non-monetary mechanism. In addition to theoretical analysis, we conduct extensive experiments using real-world datasets to generalize our routing graphs and validate the close-to-optimal performance of UPR mechanism.

In modern resource-sharing systems, multiple agents access limited resources with unknown stochastic conditions to perform tasks. When multiple agents access the same resource (arm) simultaneously, they compete for successful usage, leading to contention and reduced rewards. This motivates our theoretical study of competitive multi-armed bandit (CMAB) games. In this paper, we study a new$N$-player$K$-arm competitive MAB game, where non-myopic players (agents) compete with each other to form diverse private estimations of unknown arms over time. Their possible collisions on the same arms and the time-varying nature of arm rewards make the policy analysis here more involved than the existing studies for myopic players. We explicitly analyze the threshold-based structures of the social optimum and the existing selfish policy, showing that the latter causes prolonged convergence times$\Omega (\frac{K}{\eta ^{2}}\ln ({\frac{KN}{\delta }}))$, while the socially optimal policy with coordinated communication reduces it to$\mathcal {O}(\frac{K}{N\eta ^{2}}\ln {(\frac{K}{\delta })})$. Based on the policy comparison, we prove that the competition among selfish players for the best arm can result in an infinite price of anarchy (PoA), indicating an arbitrarily large efficiency loss compared to the social optimum. We further prove that no informational (non-monetary) mechanism (including Bayesian persuasion) can reduce the infinite PoA, as strategic misreporting by non-myopic players undermines such approaches. To address this, we propose a Combined Informational and Side-Payment (CISP) mechanism, which provides socially optimal arm recommendations with proper informational and monetary incentives to players according to their diverse and time-varying private beliefs. Our CISP mechanism keeps ex-post budget balanced for the social planner and ensures truthful reporting from players, thereby achieving the minimum$\text{PoA}=1$and the same convergence time as the social optimum.

In congestion games, selfish users behave myopically to crowd to the shortest paths, and the social planner designs mechanisms to regulate such selfish routing through information or payment incentives. However, such mechanism design requires the knowledge of time-varying traffic conditions and it is the users themselves to learn and report past road experiences to the social planner (e.g., Waze or Google Maps). When congestion games meet mobile crowdsourcing, it is critical to incentivize selfish users to explore non-shortest paths in the best exploitation-exploration trade-off. First, we consider a simple but fundamental parallel routing network with one deterministic path and multiple stochastic paths for users with an average arrival probability$\lambda $. We prove that the current myopic routing policy (widely used in Waze and Google Maps) misses both exploration (when strong hazard belief) and exploitation (when weak hazard belief) as compared to the social optimum. Due to the myopic policy’s under-exploration, we prove that the caused price of anarchy (PoA) is larger than$\frac {1}{1-\rho ^{\frac {1}{\lambda }}}$, which can be arbitrarily large as discount factor$\rho \rightarrow 1$. To mitigate such huge efficiency loss, we propose a novel selective information disclosure (SID) mechanism: we only reveal the latest traffic information to users when they intend to over-explore stochastic paths upon arrival, while hiding such information when they want to under-explore. We prove that our mechanism successfully reduces PoA to be less than 2. Besides the parallel routing network, we further extend our mechanism and PoA results to any linear path graphs with multiple intermediate nodes. In addition to the worst-case performance evaluation, we conduct extensive simulations with both synthetic and real transportation datasets to demonstrate the close-to-optimal average-case performance of our SID mechanism.

Today mobile users learn and share their traffic observations via crowdsourcing platforms (e.g., Google Maps and Waze). Yet such platforms myopically recommend the currently shortest path to users, and selfish users are unwilling to travel to longer paths of varying traffic conditions to explore. Prior studies focus on one-shot congestion games without information learning, while our work studies how users learn and alter traffic conditions on stochastic paths in a distributed manner. Our analysis shows that, as compared to the social optimum in minimizing the long-term social cost via optimal exploration-exploitation tradeoff, the myopic routing policy leads to severe under-exploration of stochastic paths with the price of anarchy (PoA) greater than 2. Besides, it fails to ensure the correct learning convergence about users' traffic hazard beliefs. To mitigate the efficiency loss, we first show that existing information-hiding mechanisms and deterministic path-recommendation mechanisms in Bayesian persuasion literature do not work with even$\mathbf{PoA}=\infty$. Accordingly, we propose a new combined hiding and probabilistic recommendation (CHAR) mechanism to hide all information from a selected user group and provide state-dependent probabilistic recommendations to the other user group. Our CHAR successfully ensures PoA less than$\frac{5}{4}$, which cannot be further reduced by any other informational mechanism. Additionally, we experiment with real-world data to verify our CHAR's good average performance.

Today mobile users learn and share their traffic observations via crowdsourcing platforms (e.g., Google Maps and Waze). Yet such platforms simply cater to selfish users' myopic interests to recommend the shortest path, and do not encourage enough users to travel and learn other paths for future others. Prior studies focus on one-shot congestion games without considering users' information learning, while our work studies how users learn and alter traffic conditions on stochastic paths in a human-in-the-loop manner. In a typical parallel routing network with one deterministic path and multiple stochastic paths, our analysis shows that the myopic routing policy (used by Google Maps and Waze) leads to severe under-exploration of stochastic paths. This results in a price of anarchy (PoA) greater than 2, as compared to the socially optimal policy achieved through optimal exploration-exploitation tradeoff in minimizing the long-term social cost. Besides, the myopic policy fails to ensure the correct learning convergence about users' traffic hazard beliefs. To address this, we focus on informational (non-monetary) mechanisms as they are easier to implement than pricing. We first show that existing information-hiding mechanisms and deterministic path-recommendation mechanisms in Bayesian persuasion literature do not work with even$\text{PoA}=\infty$. Accordingly, we propose a new combined hiding and probabilistic recommendation (CHAR) mechanism to hide all information from a selected user group and provide state-dependent probabilistic recommendations to the other user group. Our CHAR mechanism successfully ensures PoA less than$\frac{5}{4}$, which cannot be further reduced by any other informational (non-monetary) mechanism. Besides the parallel network, we further extend our analysis and CHAR mechanism to more general linear path graphs with multiple intermediate nodes, and we prove that the PoA results remain unchanged. Additionally, we carry out experiments with real-world datasets to further extend our routing graphs and verify the close-to-optimal performance of our CHAR mechanism.

In congestion games, users make myopic routing decisions to jam each other, and the social planner with the full information designs mechanisms on information or payment side to regulate. However, it is difficult to obtain time-varying traffic conditions, and emerging crowdsourcing platforms (e.g., Waze and Google Maps) provide a convenient way for mobile users travelling on the paths to learn and share the traffic conditions over time. When congestion games meet mobile crowdsourcing, it is critical to incentive selfish users to change their myopic routing policy and reach the best exploitation-exploration trade-off. By considering a simple but fundamental parallel routing network with one deterministic path and multiple stochastic paths for atomic users, we prove that the myopic routing policy's price of anarchy (PoA) can be arbitrarily large as the discount factor approaches 1. To remedy such huge efficiency loss, we propose a selective information disclosure (SID) mechanism: we only reveal the latest traffic information to users when they intend to over-explore the stochastic paths, while hiding such information when they want to under-explore. We prove that our mechanism reduces PoA to less than 2. Besides the worst-case performance, we further examine our mechanism's average-case performance by using extensive simulations.