Learning As Reinforcement: Applying Principles Of Neuroscience For More General Reinforcement Learning Agents

Abstract

A significant challenge in developing AI that can generalize well is designing agents that learn about their world without being told what to learn, and apply that learning to challenges with sparse rewards. Moreover, most traditional reinforcement learning approaches explicitly separate learning and decision making in a way that does not correspond to biological learning. We implement an architecture founded in principles of experimental neuroscience, by combining computationally efficient abstractions of biological algorithms. Our approach is inspired by research on spike-timing dependent plasticity, the transition between short and long term memory, and the role of various neurotransmitters in rewarding curiosity. The Neurons-in-a-Box architecture can learn in a wholly generalizable manner, and demonstrates an efficient way to build and apply representations without explicitly optimizing over a set of criteria or actions. We find it performs well in many environments including OpenA

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyzelikman2020learning

Related papers

• Giving Up Control: Neurons As Reinforcement Learning Agents (2020)0.00
• A Learning Gap Between Neuroscience And Reinforcement Learning (2021)0.00
• Reinforcement Learning With A Network Of Spiking Agents (2019)0.00
• An Introduction To Reinforcement Learning For Neuroscience (2023)5.84
• Reinforcement Learning And Its Connections With Neuroscience And Psychology (2020)12.25
• Learning To Act Through Evolution Of Neural Diversity In Random Neural Networks (2023)2.26
• Reinforcement Learning With Brain-inspired Modulation Can Improve Adaptation To Environmental Changes (2022)0.00
• Neuromorphic Dreaming: A Pathway To Efficient Learning In Artificial Agents (2024)0.00