Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble
Dec 6, 2021
Speakers
Seungyong Moon
Speaker · 0 followers
Jang-Hyun Kim
Speaker · 0 followers
Hyun Oh Song
Speaker · 0 followers
About
Offline reinforcement learning (offline RL), which aims to find an optimal policy from a previously collected static dataset, bears algorithmic difficulties due to function approximation errors from out-of-distribution (OOD) data points. To this end, offline RL algorithms adopt either a constraint or a penalty term that explicitly guides the policy to stay close to the given dataset. However, prior methods typically require accurate estimation of the behavior policy or sampling from OOD data points, which themselves can be a non-trivial problem. Moreover, these methods under-utilize the generalization ability of deep neural networks and often fall into suboptimal solutions too close to the given dataset. In this work, we propose an uncertainty-based offline RL method that takes into account the confidence of the Q-value prediction and does not require any estimation or sampling of the data distribution. We show that the clipped Q-learning, a technique widely used in online RL, can be leveraged to successfully penalize OOD data points with high prediction uncertainties. Surprisingly, we find that it is possible to substantially outperform existing offline RL methods on various tasks by simply increasing the number of Q-networks along with the clipped Q-learning. Based on this observation, we propose an ensemble-diversified actor-critic algorithm that reduces the number of required ensemble networks down to a tenth compared to the naive ensemble while achieving state-of-the-art performance on most of the D4RL benchmarks considered.Offline reinforcement learning (offline RL), which aims to find an optimal policy from a previously collected static dataset, bears algorithmic difficulties due to function approximation errors from out-of-distribution (OOD) data points. To this end, offline RL algorithms adopt either a constraint or a penalty term that explicitly guides the policy to stay close to the given dataset. However, prior methods typically require accurate estimation of the behavior policy or sampling from OOD data poi…
Organizer
NeurIPS 2021
Account · 1.9k followers
About NeurIPS 2021
Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Next-Generation Datasets for Safe Self-driving
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Long-Short Transformer: Efficient Transformers for Language and Vision
Watch later
Chen Zhu, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Automatic Symmetry Discovery with Lie Algebra Convolutional Network
Watch later
Nima Dehmamy, …
Total of 1 viewers voted for saving the presentation to eternal vault which is 0.1%
AutoDC: Automated data-centric processing
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Directed Spectral Measures Improve Latent Network Models Of Neural Populations
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Continuous Mean-Covariance Bandits
Watch later
Yihan Du, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%