Memory-augmented Neural Agents for Deep Model-free Reinforcement Learning
ISEF · 2020 Robotics and Intelligent Machines
Overview
While deep reinforcement learning has achieved some impressive results in the past years, it suffers from poor sample efficiency and an inablily to generalize, since they learn to manipulate specific data distributions and cannot extract principles from them. Although at its core, RL takes inspiration from the way humans learn and interact with the world, current algorithms do not reflect the human way of thinking. Memory-augmented neural networks can potentially solve these problems - the Differentiable Neural Computer represents a model of the hippocampus by accounting that humans have long and short term memory, as well as fixed routines. Furthermore, the DNC is designed to decouple learning and remembering data, which should substantially aid generalization. In this paper, we show preliminary results that the DNC exhibits great potential in the field, improving both sample efficiency and robustness to noise. Moreover, the rigid memory structures of the architecture allow for extensive analysis of the data structures stored there, which would provide invaluable novel intuition and knowledge into the processes, occurring inside the black box that is the neural network.
Competition history
- ISEF 2020
Resources
Related projects
ISEF · 2018
Optimizing Reinforcement Learning Through Dynamic Environment Manipulation
ISEF · 2023
BrainTrain: Aligning Deep Neural Networks to Human Behavior to Improve Robustness and Generalization
ISEF · 2019
Novel Reinforcement Learning Methods in Collaborative Environments
ISEF · 2017
Adaptive Variational Memory Encoding for Recurrent Data Synthesis
CWSF · 2026
From Single-Task Imitation to Real-World Generalization: A Hierarchical Reason-Act Robot System
ISEF · 2026
DOTBot: Diffusion-Aligned Chain-of-Thought Bipedal Humanoid Robot for Interpretable Real-Time Autonomous Navigation, Hazard Mitigation, and Recovery in High-Risk Built Environments
ISEF · 2023
REPNET: A Reproductive Training Architecture for Reinforcement Learning Neural Networks
ISEF · 2025
Navigating Parameter Space: Mitigating Catastrophic Forgetting in Continual Learning
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: Regeneron International Science and Engineering Fair