The Dormant Neuron Phenomenon in Deep Reinforcement Learning.
Ghada SokarRishabh AgarwalPablo Samuel CastroUtku EvciPublished in: ICML (2023)
Keyphrases
- reinforcement learning
- neural network
- function approximation
- machine learning
- learning algorithm
- state space
- markov decision processes
- reinforcement learning algorithms
- model free
- dynamic programming
- optimal policy
- multi agent reinforcement learning
- neuron model
- real time
- transition model
- function approximators
- markov decision process
- deep learning
- temporal difference
- learning agents
- activation function
- transfer learning
- policy search
- single neuron
- supervised learning