Chen Liu

Chen Liu

PhD Student in Computer Science

Krishnaswamy Lab, Yale University


Hi, my name is Chen Liu.
Or alternatively, 刘 晨 (family name followed by given name).

I am a second year CS PhD student at Yale.
My area of research is machine learning, and more specifically, manifold learning.

My research explores both theory and application of machine learning. On the research side, I focus on helping neural networks learn better representations in the latent space. On the application side, I extend my research to medical imaging and other biomedical data.

Interests
  • Manifold Learning
  • Machine Learning in Healthcare
  • Computer Vision
  • Medical Imaging
  • Computational Biology
Education

  • PhD in CS, 2022 - 2028

    Yale University

  • MS in EE, 2018 - 2019

    Columbia University

  • BS in EE, 2014 - 2018

    Bucknell University

  • Middle & High School, 2007 - 2014

    Shanghai Foreign Language School

News


[Jun 2023] Our paper has been accepted to ICML 2023 Workshop on TAG-ML. Many thanks to collaborators from Mila and Meta AI (FAIR).
[Jun 2023] Two patents came online. Kudos to my great colleagues at GE Healthcare!
[Aug 2022] Started my PhD program at Krishnaswamy Lab, Yale University.
[Jul 2022] Recognized as an Outstanding Reviewer at ICML 2022!

Awards and Recognitions

US Patent: System and Method for Obtaining Accurate Measurements and Quantification of X-Ray Image from Estimation of Key Anatomical Locations
Jun 2023
US Patent: X-Ray Lead Marker Detection System for X-Ray Imaging System
Jun 2023
Outstanding Reviewer at ICML 2022
International Conference on Machine Learning (ICML) Jul 2022
Among top 10% of reviewers at ICML 2022.
Invited as an ICML 2022 Session Chair.

Academic Service

Journal Reviewer

  1. IEEE TNNLS 2021-2023

Conference Program Committee Member

  1. NeurIPS 2021,2022,2023
  2. ICLR 2022,2023,2024
  3. ICML 2022

Teaching Fellow

  1. [Fall 2023] CPSC488 AI Foundation Models with Prof. Arman Cohan
  2. [Fall 2022] CPSC483 Deep Learning on Graph-Structured Data with Prof. Rex Ying

Experience

 
 
 
 
 
GE Healthcare
Senior Research Scientist
GE Healthcare
Aug 2021 – Jul 2022 California
 
 
 
 
 
Matician Inc
Research Software Engineer
Matician Inc
Jan 2021 – Jun 2021 California
 
 
 
 
 
Columbia University (Medical Center)
Research Assistant (Fully funded by Grant)
Columbia University (Medical Center)
Dec 2019 – Nov 2020 New York

Preprints

Quickly discover relevant content by filtering publications.
CUTS: A Framework for Multigranular Unsupervised Medical Image Segmentation
An unsupervised framework for multiscale medical image segmentation, leveraging intra-image contrastive learning, patch-level reconstruction and diffusion condensation.
PDF Cite Code ArXiv
CUTS: A Framework for Multigranular Unsupervised Medical Image Segmentation
Adversarial Focal Loss: Asking Your Discriminator for Hard Examples
A generalizable adaptation of Focal Loss to keypoint detection leveraging difficulty scores from a discriminator.
PDF Cite ArXiv
Adversarial Focal Loss: Asking Your Discriminator for Hard Examples

Journal Publications

Quickly discover relevant content by filtering publications.
Deep Learning of MRI Contrast Enhancement for Mapping Cerebral Blood Volume from Single-Modal Non-Contrast Scans of Aging and Alzheimer’s Disease Brains
A deep learning algorithm to produce Gadolinium contrast in brain MRI directly from a single non-contrast structural MRI. Reasonable prediction results confirmed by downstream scientific findings on two species, multiple studies, and various levels of brain disorders.
PDF Cite Frontiers NIH PubMed
Deep Learning of MRI Contrast Enhancement for Mapping Cerebral Blood Volume from Single-Modal Non-Contrast Scans of Aging and Alzheimer's Disease Brains
Reduced Hippocampal GABA is associated with Poorer Episodic Memory in Healthy Older Women: A Pilot Study
This is a paper describing a scientific research that is largely helped by JET.
PDF Cite Frontiers NIH PubMed
Reduced Hippocampal GABA is associated with Poorer Episodic Memory in Healthy Older Women: A Pilot Study
In vivo γ‐aminobutyric acid increase as a biomarker of the epileptogenic zone: An unbiased metabolomics approach
This is a paper describing a scientific research that is largely helped by JET.
PDF Cite Wiley Commentary on Epilepsy Currents
In vivo γ‐aminobutyric acid increase as a biomarker of the epileptogenic zone: An unbiased metabolomics approach
Understanding and Modeling Climate Impacts on Photosynthetic Dynamics with FLUXNET Data and Neural Networks
Recurrent models identifies shortwave radiation among six climate drivers as the the most influential predicting factor of photosynthetic events.
Cite Online
Understanding and Modeling Climate Impacts on Photosynthetic Dynamics with FLUXNET Data and Neural Networks

Conference Papers

Quickly discover relevant content by filtering publications.
Segmentation with Residual Attention U-Net and an Edge-Enhancement Approach Preserves Cell Shape Features
The addition of an edge-enhancement approach improved cell segmentation results over the baseline U-Net variant, in both traditional metrics and better preservation of cell shape features.
PDF Cite IEEE ArXiv
Segmentation with Residual Attention U-Net and an Edge-Enhancement Approach Preserves Cell Shape Features
Deep Learning Identifies Neuroimaging Signatures of Alzheimer’s Disease Using Structural and Synthesized Functional MRI Data
Synthesized functional MRI can facilitate classification of Alzheimer’s Disease (AD) and identification of AD’s neuroimaging signatures.
PDF Cite IEEE Arxiv
Deep Learning Identifies Neuroimaging Signatures of Alzheimer's Disease Using Structural and Synthesized Functional MRI Data
Substituting Gadolinium in Brain MRI Using DeepContrast
We develop and optimize a deep learning algorithm to produce Gd contrast in mouse brain MRI directly from a single non-contrast structural MRI.
PDF Cite IEEE Arxiv
Substituting Gadolinium in Brain MRI Using DeepContrast

Conference Workshops

Quickly discover relevant content by filtering publications.
Assessing Neural Network Representations During Training Using Noise-Resilient Diffusion Spectral Entropy
We proposed a framework to measure the entropy and mutual information in high dimensions and is applicable to neural network representations.
PDF Cite Code Arxiv ICMLW (earlier version)
Assessing Neural Network Representations During Training Using Noise-Resilient Diffusion Spectral Entropy

Conference Abstracts

Quickly discover relevant content by filtering publications.
Deep Learning Identifies Neuroimaging Signatures of Alzheimer’s Disease Using Structural and Artificial Functional MRI Data
The ISMRM (extended abstract) version of the IEEE ISBI 2021 paper.
Cite ISMRM
Deep Learning Identifies Neuroimaging Signatures of Alzheimer's Disease Using Structural and Artificial Functional MRI Data
DL-BET-A Deep Learning Based Tool for Automatic Brain Extraction from Structural Magnetic Resonance Images in Mice
See title.
Cite ISMRM
DL-BET-A Deep Learning Based Tool for Automatic Brain Extraction from Structural Magnetic Resonance Images in Mice
In Vivo GABA Increase as a Biomarker of the Epileptogenic Zone: an Unbiased Metabolomics Approach
The ISMRM (extended abstract) version of the Epilepsia 2020 paper.
Cite ISMRM
In Vivo GABA Increase as a Biomarker of the Epileptogenic Zone: an Unbiased Metabolomics Approach
JET-A Matlab Toolkit for Automated J-Difference-Edited MR Spectra Processing of In Vivo Mouse MEGA-PRESS Study at 9.4 T
A MATLAB-based software toolkit that we re-designed, re-developed and released extending a previous work.
Cite ISMRM
JET-A Matlab Toolkit for Automated J-Difference-Edited MR Spectra Processing of In Vivo Mouse MEGA-PRESS Study at 9.4 T
Predicting Gadolinium Contrast Enhancement for Structural Lesion Analysis using DeepContrast
See title.
Cite ISMRM
Predicting Gadolinium Contrast Enhancement for Structural Lesion Analysis using DeepContrast
Substituting Gadolinium In Brain MRI Using DeepContrast – A Proof-Of-Concept Study in Mice
The ISMRM (extended abstract) version of the IEEE ISBI 2020 paper.
Cite ISMRM
Substituting Gadolinium In Brain MRI Using DeepContrast -- A Proof-Of-Concept Study in Mice
Substituting Gadolinium In Human Brain MRI Using DeepContrast
We develop and optimize a deep learning algorithm to produce Gd contrast in human brain MRI directly from a single non-contrast structural MRI, and demonstrate the predicted contrast is reasonable by showing its regional vulnerability pattern to aging over the entire cortex is highly similar to the ground truth patterns.
Cite ISMRM
Substituting Gadolinium In Human Brain MRI Using DeepContrast

Miscellaneous

Quickly discover relevant content by filtering publications.
Towards a Large-Scale Unbiased Machine Learning Benchmark for Cell Instance Segmentation: Final Report for CPSC 537 Intro to Database Systems
This is my final project for the course CPSC 437/539 Intro to Database Systems in Fall 2023.
PDF Cite Project
Towards a Large-Scale Unbiased Machine Learning Benchmark for Cell Instance Segmentation: Final Report for CPSC 537 Intro to Database Systems
A Technical Deep Dive into Drag Your GAN (DragGAN)
This is a technical blog post that digs into the technical details of DragGAN (Drag Your GAN - Interactive Point-based Manipulation on the Generative Image Manifold). Specifically, we are explaining its methods section.
PDF Cite Project
A Technical Deep Dive into Drag Your GAN (DragGAN)

Contact