About Me

Research Interests

My research is on Trustworthy Machine Learning, with a focus on debiasing large pretrained models using efficient and minimally invasive approaches that maintain downstream performance.

Bio

I am a tenure-track assistant professor of Computer Science at Worcester Polytechnic Institute (WPI). Prior to my faculty position, I was a postdoctoral associate at MIT CSAIL in the Healthy ML lab, hosted by Professor Marzyeh Ghassemi. I earned my PhD at Worcester Polytechnic Institute (WPI), advised by Professor Rundensteiner and Professor Emmanuel Agu.

Recent News

• July 25: Became a tenure-track assistant professor of Computer Science at Worcester Polytechnic Institute (WPI).
• July 25: One coathored paper on how non-clinical information shapes clinical decisions in LLMs accepted to FAccT 2025!
• May 25: Serving as an Area Chair for NeurIPS 2025.
• Jan 25: One coauthored paper on temporal label noise accepted to ICLR 2025!
• Dec 24: Presenting our work on debiasing vision-language models at NeurIPS, and coauthored work on using generative models ot mitigate spurious correlations at ML4H.
• Dec 24: Two coauthored papers accepted to AAAI 2025!
• Nov 24: Coauthored paper published @ Nature Reviews Methods Primers.
• Nov 24: Presented my work on debiasing foundation models at Google’s Theory and Practice of Foundation Model Workshop.
• Oct 24: Invited to be the Senior Area Chair for the Models and Methods track at CHIL 24.
• Oct 24: First author paper on debiasing vision language models accepted @ NeurIPS 2024.
• Sep 24: Leading a SERC Group on Bias in Clinical LLMs @MIT.
• Aug 24: Coauthored paper on categorical knowledge editing published @ ACL 24.
• July 24: Coauthored paper on auditing VLMs for societal bias accepted @ AIES 24. Congrats to first author Kimia Hamidieh and collaborators Haoran Zhang, Thomas Hartvigsen and Marzyeh Ghassemi!

Select Publications

BendVLM: Test-Time Debiasing of Vision-Language Embeddings
Walter Gerych, Haoran Zhang, Kimia Hamidieh, Eileen Pan, Maanas Sharma, Thomas Hartvigsen, Marzyeh Ghassemi.
(To Appear in) NeurIPS, 2024.

MaskMedPaint: Masked Medical Image Inpainting with Diffusion Models for Mitigation of Spurious Correlations
Qixuan Jin, Walter Gerych, Marzyeh Ghassemi.
ML4H Findings Track, 2024.

Identifying implicit social biases in vision-language models.
Kimia Hamidieh, Haoran Zhang, Walter Gerych, Thomas Hartvigsen, Marzyeh Ghassemi.
AIES, 2024.

TAXI: Evaluating Categorical Knowledge Editing for Language Models.
Derek Powell, Walter Gerych, Thomas Hartvigsen.
ACL, 2024.

Learning from Time Series under Temporal Label Noise.
Sujay Nagaraj, Walter Gerych, Sana Tonekaboni, Anna Goldenberg,Berk Ustun, Thomas Hartvigsen.
Preprint, 2024.

Who Knows the Answer? Finding the Best Model & Prompt Using Confidence-Based Search.
Walter Gerych, Yara Rizk, Vatche Isahagian, Vinod Muthusamy, Evelyn Duesterwald, Praveen Venkateswaran.
AAAI, 2024.

Amalgamating Multi-Task Models with Heterogeneous Architectures.
Jidapa Thadajarassiri, Walter Gerych, Xiangnan Kong, Elke Rundensteiner.
AAAI, 2024.

Debiasing Pretrained Generative Models By Uniformly Sampling Semantic Attributes.
Walter Gerych, Kevin Hickey, Luke Buquicchio, Kavin Chandrasekaran, Abdulaziz Alajaji, Emmanuel Agu, Elke Rundensteiner.
NeurIPS, 2023.

Stabilizing Adversarial Training for Generative Networks.
Walter Gerych, Kevin Hickey* (Joint First Author), Thomas Hartvigsen, Luke Buquicchio, Abdulaziz Alajaji, Kavin Chandrasekaran, Hamid Mansoor, Emmanuel Agu, Elke Rundensteiner.
IEEE Big Data MLDB, 2023.

Knowledge Amalgamation for Multi-Label Classification via Label Dependency Transfer.
Jidapa Thadajarassiri, Thomas Hartvigsen, Walter Gerych, Xiangnan Kong, Elke Rundensteiner.
AAAI, 2023. Oral spotlight.

HAR-CTGAN: A Mobile Sensor Data Generation Tool for Human Activity Recognition.
Joshua DeOliveira, Walter Gerych, Aruzhan Koshkarova, Elke Rundensteiner, Emmanuel Agu.
IEEE Big Data 4th Special Session on HealthCare Data, 2022.

Recovering The Propensity Score From Biased Positive Unlabeled Data.
Walter Gerych, Thomas Hartvigsen, Luke Buquicchio, Emmanuel Agu, Elke Rundensteiner.
AAAI, 2022. Oral spotlight.

Robust Recurrent Classifier Chains for Multi-Label Learning with Missing Labels.
Walter Gerych, Thomas Hartvigsen, Luke Buquicchio, Emmanuel Agu, Elke Rundensteiner.
CIKM, 2022.

Stop&Hop: Early Classification of Irregular Time Series.
Thomas Hartvigsen, Walter Gerych, Jidapa Thadajarassiri, Xiangnan Kong, Elke Rundensteiner.
CIKM, 2022.

Positive Unlabeled Learning with a Sequential Selection Bias.
Walter Gerych, Thomas Hartvigsen, Luke Buquicchio, Emmanuel Agu, Elke Rundensteiner.
SDM, 2022.

Recurrent Bayesian Classifier Chains for Exact Multi-Label Classification.
Walter Gerych, Tom Hartvigsen, Luke Buquicchio, Emmanuel Agu, Elke Rundensteiner.
NeurIPS, 2021.

Misc

In my free time I enjoy visiting new places, making clothes, painting, 3D printing, and trying to keep my houseplants alive.