A Neuro-Symbolic Approach to Multi-Agent RL for Interpretability and Probabilistic Decision Making
Chitra Subramanian, Miao Liu, Naweed Khan, Jonathan Lenchner, Aporva Amarnath, Sarathkrishna Swaminathan, Ryan Riegel, Alexander Gray
TL;DR
This work tackles the interpretability and uncertainty challenges of multi-agent reinforcement learning for runtime resource management by marrying neural learning with symbolic logic. It introduces an event-driven MPOMDP framework and leverages Logical Neural Networks (LNN) for interpretable rule learning, augmented by Probabilistic Logical Neural Networks (PLNN) for probabilistic inference under partial observability. The key contributions include a formal ED-MPOMDP formulation for power sharing in Heterogeneous System-on-Chip environments, LNN-based Phase 1 rule learning with domain knowledge and guard rails, and a PLNN-based dynamic decision-making pipeline that adapts rules in real time. The results show that LNN rules improve DAG completion times under moderate to heavy load, while PLNN enables robust, probabilistic adaptation to unseen or partially observed states, yielding performance close to ideal targets and offering interpretable diagnostics for runtime control.
Abstract
Multi-agent reinforcement learning (MARL) is well-suited for runtime decision-making in optimizing the performance of systems where multiple agents coexist and compete for shared resources. However, applying common deep learning-based MARL solutions to real-world problems suffers from issues of interpretability, sample efficiency, partial observability, etc. To address these challenges, we present an event-driven formulation, where decision-making is handled by distributed co-operative MARL agents using neuro-symbolic methods. The recently introduced neuro-symbolic Logical Neural Networks (LNN) framework serves as a function approximator for the RL, to train a rules-based policy that is both logical and interpretable by construction. To enable decision-making under uncertainty and partial observability, we developed a novel probabilistic neuro-symbolic framework, Probabilistic Logical Neural Networks (PLNN), which combines the capabilities of logical reasoning with probabilistic graphical models. In PLNN, the upward/downward inference strategy, inherited from LNN, is coupled with belief bounds by setting the activation function for the logical operator associated with each neural network node to a probability-respecting generalization of the Fréchet inequalities. These PLNN nodes form the unifying element that combines probabilistic logic and Bayes Nets, permitting inference for variables with unobserved states. We demonstrate our contributions by addressing key MARL challenges for power sharing in a system-on-chip application.