Dtl: Disentangled transfer learning for visual recognition

M Fu, K Zhu, J Wu - Proceedings of the AAAI Conference on Artificial …, 2024‏ - ojs.aaai.org
When pre-trained models become rapidly larger, the cost of fine-tuning on downstream tasks
steadily increases, too. To economically fine-tune these models, parameter-efficient transfer …

Diffult: How to make diffusion model useful for long-tail recognition

J Shao, K Zhu, H Zhang, J Wu - arxiv preprint arxiv:2403.05170, 2024‏ - arxiv.org
This paper proposes a new pipeline for long-tail (LT) recognition. Instead of re-weighting or
re-sampling, we utilize the long-tailed dataset itself to generate a balanced proxy that can be …

DiffuLT: Diffusion for Long-tail Recognition Without External Knowledge

J Shao, K Zhu, H Zhang, J Wu - Advances in Neural …, 2025‏ - proceedings.neurips.cc
This paper introduces a novel pipeline for long-tail (LT) recognition that diverges from
conventional strategies. Instead, it leverages the long-tailed dataset itself to generate a …

Coarse is better? a new pipeline towards self-supervised learning with uncurated images

K Zhu, YY He, J Wu - Pattern Recognition, 2025‏ - Elsevier
Most self-supervised learning (SSL) methods often work on curated datasets where the
object-centric assumption holds. This assumption breaks down in uncurated images …