Jingyuan LI Personal Website

About Me

I am a PhD student at NeuroAI Lab, University of Washihngton , Seattle, advised by Prof.Eli Shlizerman. Before I join the University of Washington, I worked with Prof.Tai Sing Lee at Carnegie Mellon University (CMU).

My primary interest lies in general understanding of time-series data, particular to time-series neural signals and time-series behavior recordings.
In terms of method, I focus on deep learning based method, such as Recurrent Neural Network, Graph Neural Network and Transformer.
My research topics include Semi(Un)-supervised Learning, Representation Learning, Computational Neuroscience, Brain Computer Interface, Robust and Uncertainty Evaluation.

Publications

Brain-to-Text Decoding with Context-Aware Neural Representations and Large Language Models

Authors: Jingyuan Li, Trung Le, Chaofei Fan, Mingfei Chen, Eli Shlizerman.
In submission to ICLR 2025

I developed DCoND-LIFT, a method that leverages large language models to translate neural signals into text, reducing the word error rate from 9.93% to 5.77% compared to the baseline.

Keywords:Brain-Computer Interface, Brain-to-Text, Large Language Model, Phoneme Decoding

Translating Mental Imaginations into Characters with Codebooks and Dynamics-Enhanced Decoding

Authors: Jingyuan Li,Yansen Wang, Nie Lin, Dongsheng Li.
In submission to ICASSP 2025
Preprint

I introduced four mental tasks that map EEG recordings to 35 characters, including letters and numbers, with high accuracy, significantly improving over direct character imagination by aounrd 8.5%.

Keywords: EEG, Brain-Computer Interface, Brain-to-Character, GRU, CNN

Self-supervised behavior modeling with dense keypoint tracking

Authors: Ying Yu*, Jingyuan Li*, Kun Su, Anna Bowen, Carlos Campos. *Equal contribution.
Cosyne 2024 Oral
Talk

I presented 2D to 3D mapping of animal posture, from which behavior segment can be retrieved with one-shot example.

Keywords: Posture modeling, Behavior segmentation

AMAG: Additive, Multiplicative and Adaptive Graph Neural Network For Forecasting Neural Activity

Authors: Jingyuan Li, Leo Scholl, Trung Le, Pavithra Rajeswaran, Amy Orsborn, Eli Shlizerman.
NeurIPS 2023
Paper

I introduce a Graph Neural Network for predicting future neural signals, which helps to recover the underlying neural dependencies.

Keywords: intracortical recordings, Brain-Computer Interface, Graph Neural Networks

AL-SAR: Active Learning for Skeleton-based Action Recognition

Authors: Jingyuan Li, Trung Le, Eli Shlizerman.
IEEE TNNLS, 2023
Paper

I proposed an active learning method for semi-supervised behavior recognition, achieving strong action recognition performance under the circumstances when only a limited number of annotated samples are available.

Keywords: Active Learning, semi-supervised Learning, Action Recognition

OpenLabCluster: Active Learning Based Clustering and Classification of Animal Behaviors in Videos Based on Automatically Extracted Kinematic Body Keypoints

Authors: Jingyuan Li, Moishe Keselman, Eli Shlizerman.
arXiv, 2022
Preprint GitHub

I implemented an active learning pipeline for animal behavior recognition requiring a few of labeled animal behavior examples, a GUI was introduced for general uses.

Keywords: Animal Behavior Recognition, Active Learning, Graphic User Interface

Hello, I'm Jingyuan Li