Neuro-symbolic approaches have been proposed to improve deep reinforcement learning (RL), for instance, utilizing external knowledge or directly explaining what is learned.
As an example, Logical Neural Networks (LNN) simultaneously provide key properties of both neural networks (learning) and symbolic logic (reasoning), thus providing us with the automated possibility to find logical optimal solutions.
A recent paper, published on arXiv.org, presents a Logical Optimal Actions architecture for neuro-symbolic RL applications with LNN for text-based interaction games.
The proposed demonstration uses a learning environment as a miniature of a natural language-based interactive environment. It provides a web-based user interface for visualizing the game interaction. Trained and pre-defined logical rules are also visualized to help the human user understand the benefits of introducing the logical rules via neuro-symbolic frameworks.
We present Logical Optimal Actions (LOA), an action decision architecture of reinforcement learning applications with a neuro-symbolic framework which is a combination of neural network and symbolic knowledge acquisition approach for natural language interaction games. The demonstration for LOA experiments consists of a web-based interactive platform for text-based games and visualization for acquired knowledge for improving interpretability for trained rules. This demonstration also provides a comparison module with other neuro-symbolic approaches as well as non-symbolic state-of-the-art agent models on the same text-based games. Our LOA also provides open-sourced implementation in Python for the reinforcement learning environment to facilitate an experiment for studying neuro-symbolic agents. Code: this https URL
Research paper: Kimura, D., “LOA: Logical Optimal Actions for Text-based Interaction Games”, 2021. Link: https://arxiv.org/abs/2110.10973
Discover more from Today Headline
Subscribe to get the latest posts to your email.