Gregory Kondas

Gregory Kondas

Research Assistant @ Michigan AI Lab

About Me

Hello! I am a full-time research assistant at the Michigan AI lab working with Professor Jenna Wiens. I am fortunate to have Sarah Jabbour as an awesome mentor! My research focuses on machine learning with applications in healthcare. I am currently interested in foundation models for electronic health records and wearable data, as well as medical time series analysis and explainable AI. I will soon be applying to Ph.D. programs with start dates in the fall of 2025. Here is my CV.

I graduated from the University of Michigan in 2024 with a B.S. in Computer Science and a minor in Physics (Go Blue!). I also completed ample coursework in the life sciences. In the past, my broad interest in science led me to a variety of research fields. I worked on problems in computational neuroscience with Professor Thad Polk. I also spent a summer optimizing the construction and testing of components for CERN’s upgraded muon spectrometer under the direction of Professor Bing Zhou.

Outside of research I enjoy basketball, keeping up with current events, music, and cooking!

Email: gkondas (at) umich (dot) edu

News

  • 11/24: I wrote a blog post on DEPICT for the Michigan AI lab blog. Check it out here.
  • 7/24: DEPICT: Diffusion-Enabled Permutation Importance for Image Classification Tasks accepted to ECCV 2024!
  • 6/24: Started as a full time research assistant with Jenna Wiens’s MLD3 group at Michigan.
  • 5/24: Graduated with a B.S. in Computer Science and minor in Physics from Michigan with distinction.
  • 6/23: Started as a undergraduate research assistant with Jenna Wiens’s MLD3 group at Michigan.

Publications

DEPICT: Diffusion Enabled Permutation Importance for Image Classification Tasks
Sarah Jabbour, Gregory Kondas, Ella Kazerooni, Michael W. Sjoding, David Fouhey*, Jenna Wiens*. ECCV 2024. [paper][code]
depict_thumbnail Permutation importance has been used to provide feature importance explanations for tabular-based models. Leveraging text-conditioned diffusion, we extend this framework to image-based models and facilitate dataset-level model explanations.

Teaching

  • Teaching Assitstant - PHYSICS 250: Physics for the Life Sciences II, Spring 2021 - Fall 2023