Posted:
Aug 7, 2023 ·
Updated: Aug 14, 2023 · 4 min.
In paper SeLa (Self-labeling), the cluster-based self-supervised method is transformed to a transportation problem, and optimized using Sinkhorn-Knopp algorithm. This paper refined the methods by using contrastive learning between views and applied multi-crop technique.
Posted:
Aug 3, 2023 ·
Updated: Nov 16, 2023 · 8 min.
OpenMatch introduces a detector to filter out outliers from unlabeled data, and then learn like FixMatch. Its representation is also adapted to the task.
Posted:
Jul 28, 2023 ·
Updated: Jul 29, 2023 · 6 min.
Basic notion in reinforce learning. Agent, state, action, policy, reward, state transition and value function.
Posted:
Jul 28, 2023 ·
Updated: Jul 29, 2023 · 6 min.
A brief guide to post template.