Publications
Selected papers and preprints. A curated list of work on representation learning and reinforcement learning. For citation counts and the most current index, visit Google Scholar or DBLP .
All Conference Journal Preprints
2024
Conference · ICLR 2024
State Representation Learning Using an Unbalanced Atlas Li Meng, Morten Goodwin, Anis Yazidi, Paal E. Engelstad
A manifold-based self-supervised learning paradigm that scales state representations to high-dimensional target spaces.
Self-supervised learning Manifold learning Reinforcement learning
Preprint · arXiv preprint
Maximum Manifold Capacity Representations in State Representation Learning Li Meng, Morten Goodwin, Anis Yazidi, Paal Engelstad
Extends manifold-aware state representation learning with a nuclear-norm objective and a more efficient route to manifold-capacity regularization.
Manifold capacity Representation learning Atari
Preprint · arXiv preprint
A Manifold Representation of the Key in Vision Transformers Li Meng, Morten Goodwin, Anis Yazidi, Paal Engelstad
Studies decoupled query–key–value construction and introduces manifold structure into the key representation of vision transformers.
Vision transformers Manifold learning Attention
2023
Conference · CAIP 2023
Unsupervised State Representation Learning in Partially Observable Atari Games Li Meng, Morten Goodwin, Anis Yazidi, Paal Engelstad
Develops an unsupervised state representation scheme for partially observable environments and evaluates it on Atari-based benchmarks.
Partial observability State representation Contrastive learning
2022
Journal · IEEE Transactions on Games
Improving the Diversity of Bootstrapped DQN by Replacing Priors With Noise Li Meng, Morten Goodwin, Anis Yazidi, Paal Engelstad
Replaces fixed randomized priors with Gaussian noise to sustain ensemble diversity and improve deep exploration in Atari games.
Deep reinforcement learning Exploration DQN
Conference · NLDL 2022
Expert Q-learning: Deep Reinforcement Learning with Coarse State Values from Offline Expert Examples Li Meng, Anis Yazidi, Morten Goodwin, Paal Engelstad
Incorporates coarse expert state-value information into deep Q-learning to guide online learning from limited demonstrations.
Expert knowledge Q-learning Offline examples
Preprint · arXiv preprint
Deep Reinforcement Learning with Swin Transformers Li Meng, Morten Goodwin, Anis Yazidi, Paal Engelstad
Investigates Swin Transformer backbones for online deep reinforcement learning across a large suite of Atari games.
Swin Transformer Deep RL Visual learning