This work identifies a process-agnostic structural condition, based on sliding-window mixing, that ensures the graph's intrinsic walk remains stable for both exploration and navigation and establishes sublinear expected regret.
Abstract
We study stochastic multi-armed bandits on dynamic graphs, where arms correspond to the vertices of a network with time-varying edges. In this setting, the learner is restricted to local movement, selecting only its current node or an immediate neighbor at each round. This constraint decouples best-arm identification from exploitation: even after the optimal arm is identified, the learner may remain unable to reach it through the evolving topology. We identify a process-agnostic structural condition, based on sliding-window mixing, that ensures the graph's intrinsic walk remains stable for both exploration and navigation. Under this regime, we analyze a family of local explore-then-commit algorithms and establish sublinear expected regret. Our framework includes a reward-aware strategy, for which we prove a worst-case safety theorem and a separate performance gain theorem.
The results show that residual-graph value learning yields state-dependent dynamic matching policies that adapt to realized connectivity and exit information.
In this work, we extend the Dependent Click Model (DCM) Bandits to a multiplayer information-asymmetric setting, where multiple agents interact with a shared ranked list and may observe multiple clicks per session, introducing new challenges for selection strategies. We study asymmetry in (1) actions and (2) rewards, p...
Andy Wang, Charlton Shih, William Chang· 2026 8th Asia Conference on...· 0 citations
Results show that a modest amount of predicted information is enough to accelerate decentralized routing well below Kleinberg's classical bound, and that even when nodes reveal no coordinates at all, reliable delivery remains achievable.
Ladan Kian, M. Tan, Dariusz R. Kowalski· 0 citations
Network common learning is introduced, a network analogue of common learning, and it is shown that it is attained when neighboring agents'observations differ by many signals, as on the two-dimensional grid, but fails on networks with informational bottlenecks, such as the line.
Olga Rospuskova, Omer Tamuz, Jake Zhang· 0 citations
Distributed adaptation and learning over directed, unbalanced graphs poses unique challenges due to asymmetric communication and heterogeneous data across nodes. In this work, we introduce a novel class of first-order primal–dual stochastic gradient algorithms for such graphs. Our flagship algorithm, called primal-dual...
Sheng Zhang, Hong-Yu Han, Hong-Yang Chen et al.· IEEE Transactions on Signal...· 0 citations
A three-player game is constructed with a preferentially stable set whose span is dynamically unstable, showing that preferences do not suffice as a criterion of dynamic stability and bridges the gap via the notion of resilience under aggregate deviations.
Omar Abbadi, R. Laraki, P. Mertikopoulos· 2 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
