Zhihan Zhou

Zhihan Zhou (周之涵)

I am a final year Ph.D. candidate in the Computer Science Department at Northwestern University, advised by Prof. Han Liu. I received my Bachelor's Degree in Mathematics at Zhejiang University and Master's Degree in Computer Science at Northwestern University. I did several research interns at NVIDIA, Amazon, and Snapchat.

Email / CV / Google Scholar / Linkedin / Github

Research

My research focuses on Large Language Models (LLM) and Deep Learning (DL). I mainly work on LLM for Genomics. I have also worked several projects on conversational AI and user modeling.

	USE: Dynamic User Modeling with Stateful Sequence Models Zhihan Zhou, Qixiang Fang, Leonardo Neves, Yozen Liu, Francesco Barbieri, Han Liu, Maarten W. Bos, Ron Dotsch (*: co-first author) Preprint, 2024 paper Seamlessly integrating past and present user behavior sequences for user embedding.
	DNABERT-S: Learning Species-Aware DNA Embedding with Genome Foundation Models Zhihan Zhou, Weimin Wu, Harrison Ho, Jiayi Wang, Lizhen Shi, Ramana V Davuluri, Zhong Wang, Han Liu (*: co-first author) Preprint, 2024 paper / code / model DNABERT-S is a foundation model based on DNABERT-2 specifically designed for generating DNA embedding that naturally clusters and segregates genome of different species in the embedding space.
	General-Purpose User Modeling with Behavioral Logs: A Case Study with Snapchat Qixiang Fang, Zhihan Zhou, Francesco Barbieri, Yozen Liu, Leonardo Neves, Dong Nguyen, Daniel L. Oberski, Maarten W. Bos, Ron Dotsch (*: co-first author) SIGIR 2024 paper A Transformer-based model that learns general-purpose user embedding from SnapChat user behavioral logs.
	Efficient Action Robust Reinforcement Learning with Probabilistic Policy Execution Uncertainty Guanlin Liu, Zhihan Zhou, Han Liu, Lifeng Lai Preprint, 2023 paper A novel approach for solving action robust RL problems with probabilistic policy execution uncertainty.
	DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genome Zhihan Zhou, Yanrong Ji, Weijian Li, Pratik Dutta, Ramana Davuluri, Han Liu ICLR 2024 paper / code / model We introduce DNABERT-2, an efficient and effective foundation model for multi-species genome that achieves state-of-the-art performance with 20 time less parameters. We also provide a benchmark Genome Understanding Evaluation (GUE) containing 28 datasets across 7 tasks.
	KGML-xDTD: A Knowledge Graph-based Machine Learning Framework for Drug Treatment Prediction and Mechanism Description Chunyu Ma, Zhihan Zhou, Han Liu, David Koslicki GigaScience, 2023 code Optimizing Graph Neural Networks and Reinforcement Learning for link prediction and path finding on one of the largest biomedical knowledge graphs.
	Learning Dialogue Representations from Consecutive Utterances Zhihan Zhou, Dejiao Zhang, Wei Xiao, Nicholas Dingwall, Xiaofei Ma, Andrew O Arnold, Bing Xiang NAACL 2022 paper / code Introduce DSE, a pre-trained language model that achieves great performance in few-shot dialogue understanding. Trained by contrastive learning on consecutive utterances of dialogues.
	Trade the Event: Corporate Events Detection for News-Based Event-Driven Trading Zhihan Zhou, Liqian Ma, Han Liu ACL 2021, Findings paper / video / code Analyze the impact of news articles on the stock market. Consider corporate events are the driven force and introduce methods for corporate event detection. Provide new datasets for text-based stock prediction and event detection.
	DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome Yanrong Ji, Zhihan Zhou, Han Liu, Ramana V Davuluri (: co-first author) Bioinformatics, 2021* paper / code / model Introduce new paradigm for DNA analysis. Pre-trained on human genome and achieves state-of-the-art performance in a wide range of DNA analysis problems including promoter prediction and splice prediction.
	Few-shot Slot Tagging with Collapsed Dependency Transfer and Label-enhanced Task-adaptive Projection Network Yutai Hou, Wanxiang Che, Yongkui Lai, Zhihan Zhou, Yijia Liu, Han Liu, Ting Liu ACL 2020 paper / code Introduce novel method for few-shot sequence labeling by incorporating label information and learned label transfer into few-shot prediction.
	Joint Speaker Diarization and Recognition Using Convolutional and Recurrent Neural Networks Zhihan Zhou, Yichi Zhang, Zhiyao Duan ICASSP 2018 paper Combines speaker identity predictions and speaker change rate predictions for joint speaker diarization and recognition.
	ELLPMDA: Ensemble learning and link prediction for miRNA-disease association prediction Xing Chen, Zhihan Zhou, Yan Zhao (: co-first author) RNA Biology, 2018* paper Perform ensemble learning on multiple graph learning method to achieve accurate miRNA-disease association prediction.