Provable reward-agnostic preference-based reinforcement learning

W Zhan, M Uehara, W Sun, JD Lee - ar** generally-capable agents that can
adapt to new tasks without additional training in the environment. Learning world models …

Task-prompt generalised world model in multi-environment offline reinforcement learning

X **ong, L Meng, J Ruan, Q Zhang, G Li, D **ng… - ECAI 2023, 2023 - ebooks.iospress.nl
Offline reinforcement learning (RL) circumvents costly interactions with the environment by
utilising historical trajectories. Incorporating a world model into this method could …

Task aware dreamer for task generalization in reinforcement learning

C Ying, Z Hao, X Zhou, H Su, S Liu, D Yan… - arxiv preprint arxiv …, 2023 - arxiv.org
A long-standing goal of reinforcement learning is to acquire agents that can learn on training
tasks and generalize well on unseen tasks that may share a similar dynamic but with …

Model-Based Reinforcement Learning With Isolated Imaginations

M Pan, X Zhu, Y Zheng, Y Wang… - IEEE Transactions on …, 2023 - ieeexplore.ieee.org
World models learn the consequences of actions in vision-based interactive systems.
However, in practical scenarios like autonomous driving, noncontrollable dynamics that are …

Transferable Reinforcement Learning via Generalized Occupancy Models

C Zhu, X Wang, T Han, SS Du… - ICML 2024 Workshop on …, 2024 - openreview.net
Intelligent agents must be generalists, capable of quickly adapting to various tasks. In
reinforcement learning (RL), model-based RL learns a dynamics model of the world, in …

Filtered observations for model-based multi-agent reinforcement learning

L Meng, X **ong, Y Zang, X Zhang, G Li, D **ng… - … Conference on Machine …, 2023 - Springer
Reinforcement learning (RL) pursues high sample efficiency in practical environments to
avoid costly interactions. Learning to plan with a world model in a compact latent space for …

[PDF][PDF] World model architectures in reinforcement learning: an exploration of strengths and limitations

R Schiewer - 2024 - d-nb.info
Reinforcement learning (RL), a machine learning algorithm inspired by the trial-and-error
learning mechanism of natural intelligence, is characterised by agents that interact with their …

Enhancing robustness and interpretability in natural language processing through representation learning

H Yan - 2024 - wrap.warwick.ac.uk
Recent advancements in Natural Language Processing (NLP), particularly those driven by
representation learning, have led to significant breakthroughs in a range of applications …