Hao Zou

BS Student
Computer Science Department
University of Minnesota, Twin Cities

pronouns: he/him

Hi! I'm a Senior undergraduate student in Computer Science at University of Minnesota, Twin Cities (UMN), where I work in the Minnesota NLP Group and UMN Data Analysis and Management Research Group. My advisors are Prof. Dongyeop Kang and Prof. Jaideep Srivastava.

My research interests are in Machine Learning (ML) and Natural Language Processing (NLP), particularly in its intersections with Robustness and Generalization. To better combine applications with foundational theories, I share interests in theoretical ML as well, specifically in Deep Learning (DL). Current machine learning techniques fall short of human abilities in both their capacity to learn with causal reasoning and to learn robust prediction mechanisms. My research aims to (1) understand and improve the robustness in learning, and (2) build interpretable and trustworthy systems that can communicate effectively with humans and learn from human guidance and controls. These interests and goals have drawn me toward three main research directions:
(1) Diffusion Models for compositional controllable generations and adversarial robustness;
(2) Causality in NLP;
(3) Understanding Adversarial Examples.

Since May 2022, I started working with Prof. Weijie Su and Prof. Hangfeng He at University of Pennsylvania as Research Assistant. I lead a project about A New Theory of Adversarial Examples and Generalization in Neural Networks (NN).

Since May 2021, I interned at the CLIP Lab of University of Maryland Institute for Advanced Computer Studies (UMIACS) under Prof. Jordan Boyd-Graber on improving Neural Question Answering Systems by leveraging self-supervised question generation. Previously, I've also spent time at the Signal Processing and Information Systems Laboratory, University of Rochester, working with Prof. Mujdat Cetin.

Thanks to the support of my research mentors, it's my grear fortunate and honor to begin working on research early in my undergraduate career with those awesome researchers!

Email / CV / LinkedIn / GitHub / Twitter