5 papers
Certificate-Guided Evaluation of Reinforcement Learning Generalization
Vignesh Subramanian, ÄorÄe ŽikeliÄ, Suguman Bansal
This work presents a logic-driven framework to evaluate the performance of reinforcement learning (RL) algorithms in their ability to generalize to unseen tasks. Our framework defi…
Decoupled Behavioral Cloning for Scalable Inductive Generalization in RL from Specifications
Vignesh Subramanian, Subhajit Roy, Suguman Bansal
Inductive generalization is a framework for reinforcement learning (RL) generalization in which inductively related task instances admit inductively related policies. Prior work ca…
Reinforcement Learning for Reachability: Guaranteeing Asymptotic Optimality
Amogh Palasamudram, Jakub Svoboda, Suguman Bansal +1
Reinforcement learning (RL) for reachability specifications is fundamental in sequential decision-making, yet theoretical guarantees remain less explored. A recent work achieves as…
INTERLEAVE: A Faster Symbolic Algorithm for Maximal End Component Decomposition
Suguman Bansal, Ramneet Singh
This paper presents a novel symbolic algorithm for the Maximal End Component (MEC) decomposition of a Markov Decision Process (MDP). The key idea behind our algorithm INTERLEAVE is…
Inductive Generalization in Reinforcement Learning from Specifications
Vignesh Subramanian, Rohit Kushwah, Subhajit Roy +1
We present a novel inductive generalization framework for RL from logical specifications. Many interesting tasks in RL environments have a natural inductive structure. These induct…