arXiv subjects

Amit Kiran Rege

Publications and source records attributed to Amit Kiran Rege.

9 recordsLinked to original sources

arXiv2026

Learning from Local Walks on Dynamic Graphs with Bandit Feedback

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.

cs.LG↗

arXiv2026

Flickering Multi-Armed Bandits

We introduce Flickering Multi-Armed Bandits (FMAB) to model sequential decision-making in environments with changing action availability, where accessibility of the next action is restricted to a subset dependent on the agent's current choice. We formalize these constraints through stochastically evolving graphs where actions are limited to local neighborhoods. This mobility-constrained structure imposes a dual challenge: the statistical requirement of information acquisition and the physical overhead of navigation. We analyze FMAB under i.i.d. Erdős--R'enyi and Edge-Markovian process, proposing a two-phase lazy random walk algorithm for robust exploration. We establish high-probability sublinear regret bounds and prove near-optimality via a matching information-theoretic lower bound. Our results characterize the intrinsic cost of learning under local-move constraints, complemented by a robotic disaster-response simulation.

cs.LG↗

arXiv2026

Multi-Agent Lipschitz Bandits

We study the decentralized multi-player stochastic bandit problem over a continuous, Lipschitz-structured action space where hard collisions yield zero reward. Our objective is to design a communication-free policy that maximizes collective reward, while separating coordination costs from learning costs. We propose a modular protocol that first solves the multi-agent coordination problem by identifying and seating players on distinct, high-value regions via a novel maxima-directed search and then decouples the problem into $N$ independent single-player Lipschitz bandits. In the consensus regime, we obtain an end-to-end regret bound whose dominant learning term is $\tilde{O}(T^{(d+1)/(d+2)})$, matching the single-player Lipschitz rate; the upfront coordination cost is horizon-independent at fixed confidence and only polylogarithmic in $T$ in the expected-regret form. Under an additional public coverage/scheduling assumption for the epochic extension, we also obtain a gap-free $\tilde{O}(T^{(d+1)/(d+2)})$ guarantee. We further derive a matching lower bound for the dominant learning term and extend the framework to general distance-threshold collision models.

cs.LG↗

arXiv2026

A Unified Framework for Locality in Scalable MARL

Scalable methods for networked multi-agent reinforcement learning let each agent plan using only a small neighborhood of the agent graph. This works only when the system is value-local, meaning a perturbation at one agent affects the long-run value at another agent weakly when the two are far apart. In the average-reward setting, the standard way to certify locality is the Dobrushin row-sum bound on a single matrix $C^π$ that captures how each agent's next state depends on each other agent's current state. To make this matrix easy to work with, prior work bounds it by a supremum over joint actions. The resulting bound is independent of the policy, but it is loose whenever the policy never picks the worst-case action. We split $C^π$ into pieces that separately track environment sensitivity and policy sensitivity, $C^π\preceq E^{\mathrm s}+E^{\mathrm a}Π(π)$, where $E^{\mathrm s}$ measures how the next state moves with the current state, $E^{\mathrm a}$ measures how it moves with the current action, and $Π(π)$ measures how reactive the policy is to changes in state. The spectral radius of $H^π:= E^{\mathrm s}+E^{\mathrm a}Π(π)$ then controls the decay of the average-reward Poisson solution, and the spectral certificate $ρ(H^π)<1$ is strictly weaker than the row-sum condition $\|H^π\|_\infty<1$ on the same matrix and applies in regimes where policy-independent action-supremum bounds used in prior Dobrushin-style work cannot. For temperature-$τ$ softmax policies we get $Π(π)\le L/(2τ)$, so the softmax temperature directly controls locality. We use this decay result to give a deterministic oracle guarantee for a block-coordinate KL-proximal policy-improvement template whose truncation bias decays exponentially in the message-passing radius $κ$.

cs.LG↗

arXiv2026

The Role of Generator Access in Autoregressive Post-Training

We study how generator access constrains autoregressive post-training. The central question is whether the learner is confined to fresh root-start rollouts or can return to previously built prefixes and query the next-token rule there. In the root-start regime, output sampling, generated-token log probabilities, top-$k$ reports, and full next-token distributions along sampled trajectories all reduce to one canonical experiment, limited by the on-policy probability of reaching informative prefixes. Weak prefix control breaks this barrier, and once control is available, richer observations such as conditional sampling or logits can outperform top-$1$ access. Changing only the generator interface creates an exponential gap for KL-regularized outcome-reward post-training.

cs.LG↗

arXiv2026

Data Attribution in Adaptive Learning

Machine learning models increasingly generate their own training data -- online bandits, reinforcement learning, and post-training pipelines for language models are leading examples. In these adaptive settings, a single training observation both updates the learner and shifts the distribution of future data the learner will collect. Standard attribution methods, designed for static datasets, ignore this feedback. We formalize occurrence-level attribution for finite-horizon adaptive learning via a conditional interventional target, prove that replay-side information cannot recover it in general, and identify a structural class in which the target is identified from logged data.

cs.LG↗

arXiv2025

Incentivized Lipschitz Bandits

We study incentivized exploration in multi-armed bandit (MAB) settings with infinitely many arms modeled as elements in continuous metric spaces. Unlike classical bandit models, we consider scenarios where the decision-maker (principal) incentivizes myopic agents to explore beyond their greedy choices through compensation, but with the complication of reward drift--biased feedback arising due to the incentives. We propose novel incentivized exploration algorithms that discretize the infinite arm space uniformly and demonstrate that these algorithms simultaneously achieve sublinear cumulative regret and sublinear total compensation. Specifically, we derive regret and compensation bounds of $\Tilde{O}(T^{d+1/d+2})$, with $d$ representing the covering dimension of the metric space. Furthermore, we generalize our results to contextual bandits, achieving comparable performance guarantees. We validate our theoretical findings through numerical simulations.

cs.LG↗

arXiv2025

Hamiltonian Learning using Machine Learning Models Trained with Continuous Measurements

We build upon recent work on using Machine Learning models to estimate Hamiltonian parameters using continuous weak measurement of qubits as input. We consider two settings for the training of our model: (1) supervised learning where the weak measurement training record can be labeled with known Hamiltonian parameters, and (2) unsupervised learning where no labels are available. The first has the advantage of not requiring an explicit representation of the quantum state, thus potentially scaling very favorably to larger number of qubits. The second requires the implementation of a physical model to map the Hamiltonian parameters to a measurement record, which we implement using an integrator of the physical model with a recurrent neural network to provide a model-free correction at every time step to account for small effects not captured by the physical model. We test our construction on a system of two qubits and demonstrate accurate prediction of multiple physical parameters in both the supervised and unsupervised context. We demonstrate that the model benefits from larger training sets establishing that it is in fact "learning," and we show robustness to errors in the assumed physical model by achieving accurate parameter estimation in the presence of unanticipated single particle relaxation.

quant-ph↗

arXiv2024

Where Did Your Model Learn That? Label-free Influence for Self-supervised Learning

Self-supervised learning (SSL) has revolutionized learning from large-scale unlabeled datasets, yet the intrinsic relationship between pretraining data and the learned representations remains poorly understood. Traditional supervised learning benefits from gradient-based data attribution tools like influence functions that measure the contribution of an individual data point to model predictions. However, existing definitions of influence rely on labels, making them unsuitable for SSL settings. We address this gap by introducing Influence-SSL, a novel and label-free approach for defining influence functions tailored to SSL. Our method harnesses the stability of learned representations against data augmentations to identify training examples that help explain model predictions. We provide both theoretical foundations and empirical evidence to show the utility of Influence-SSL in analyzing pre-trained SSL models. Our analysis reveals notable differences in how SSL models respond to influential data compared to supervised models. Finally, we validate the effectiveness of Influence-SSL through applications in duplicate detection, outlier identification and fairness analysis. Code is available at: \url{https://github.com/cryptonymous9/Influence-SSL}.

cs.LG↗