Beyond One-Size-Fits-All: Diagnosis-Driven Online Reinforcement Learning with Offline Priors
Let deployment evidence guide how online learners use and adapt their offline priors.
Guozheng Ma · 马国政 Reinforcement learning · NTU, Singapore
Guozheng Ma 马国政
PhD student at Nanyang Technological University,
advised by Dacheng Tao.
I study how reinforcement learning agents can scale, adapt, and keep learning in open-ended environments — from optimization and plasticity to exploration and generalization.
Greatness cannot be planned, so I play with joyful explorations.
Ideas, in a little more detail
Research notes
Sparsity & scaling
Why sparsify RL networks?
Static sparsity offers a surprisingly simple way to improve scaling in deep RL.
Plasticity & adaptation
Keeping visual RL able to learn
Data, the critic and the timing of updates all shape an agent's learning capacity.
Data & generalization
Learning from a different view
Useful visual diversity depends on both the transformation and the way it enters training.
From questions to experiments
Selected publications
Network Sparsity Unlocks the Scaling Potential of Deep Reinforcement Learning
Randomly pruning a network before training improves how deep RL scales.
Revisiting Plasticity in Visual Reinforcement Learning: Data, Modules and Training Stages
Understanding when and where visual RL loses its ability to keep learning.