Zheng HE

Contact by zhhe@cs.ubc.ca

I am a PhD track student at the Department of Computer Science at University of British Columbia, where I am lucky to be advised by Prof. Danica J. Sutherland. Currently, I am visiting Prof. Arthur Gretton’s group at Gatsby Unit, University College London, working on kernel-based statistical testing of conditional independence. I received my Master degree in 2023 and Bachelor degree in 2020, from Beihang University.

My research interests span the areas of statistical machine learning and deep learning. I am excited to delve into when and why the models and algorithms work (or do not work). And I hope to understand the remarkable successes and intriguing behaviors exhibited by these learning models better throughout my research journey.

latest posts

Dec 30, 2023	a note on learning bounds
Oct 20, 2023	a lecture note on modern statistical learning theory
Feb 1, 2023	learning with biased labels
Jul 12, 2022	sparse double descent where network pruning aggravates overfitting
Aug 30, 2021	overparameterization and sparsity

publications

Sparse double descent: Where network pruning aggravates overfitting

Zheng He, Zeke Xie, Quanzhi Zhu, and Zengchang Qin

In International Conference on Machine Learning, ICML, 2022

Abs arXiv Poster Code

People usually believe that network pruning not only reduces the computational cost of deep networks, but also prevents overfitting by decreasing model capacity. However, our work surprisingly discovers that network pruning sometimes even aggravates overfitting. We report an unexpected sparse double descent phenomenon that, as we increase model sparsity via network pruning, test performance first gets worse (due to overfitting), then gets better (due to relieved overfitting), and gets worse at last (due to forgetting useful information). While recent studies focused on the deep double descent with respect to model overparameterization, they failed to recognize that sparsity may also cause double descent. In this paper, we have three main contributions. First, we report the novel sparse double descent phenomenon through extensive experiments. Second, for this phenomenon, we propose a novel learning distance interpretation that the curve of l2 learning distance of sparse models (from initialized parameters to final parameters) may correlate with the sparse double descent curve well and reflect generalization better than minima flatness. Third, in the context of sparse double descent, a winning ticket in the lottery ticket hypothesis surprisingly may not always win.
Random Neural Graph Generation with Structure Evolution

Yuguang Zhou, Zheng He, Tao Wan, and Zengchang Qin

In Neural Information Processing: 28th International Conference, ICONIP, 2021

Abs HTML

In deep learning research, typical neural network models are multi-layered architectures, and weights are tuned while optimizing a carefully designed loss function. In recent years, studies of randomized neural networks have been extended towards deep architectures, opening a new research direction to the design of deep learning models. However, how the structure of the network can influence the model performance still remains unclear. In this paper, we move a further step to investigate the relation between network topology and performance via a structure evolution algorithm. Experimental results show that the graph would evolve towards a more small-world topology at the beginning of the training session along with gaining accuracy, and would also evolve towards a structure with more scale-free property in the following periods. These conclusions could help explain the effectiveness of the randomly connected networks, as well as give us insights in new possibilities of network architecture design.