Machine Learning for Medicine

I work with clinicians and data scientists to develop NLP methods for extracting patient information from clinical notes and social media data and for understanding the nosology of complex diseases better.

Noelle Lim, Claire S. Lee, and Michael Guerzhoy

Detecting a Proxy for Potential Comorbid ADHD in People Reporting Anxiety Symptoms from Social Media Data

CLPsych at EACL, 2024

Interpretability and Representation Learning

How do neural networks actually process information? I investigate what features neural networks learn and how they use them, from understanding positional encodings in transformers to how ConvNets process color and intensity.

Andy Cai, Sajad Hashemi, Noah Paulson, and Michael Guerzhoy

Latent Representation of Microstructures using Variational Autoencoders with Spatial Statistics Space Loss

ECLR Workshop at ICCV, 2025

Zhi (Whitney) Ji and Michael Guerzhoy

Shape-Based Features Complement CLIP Features And Features Learned from Voxels in 3D Object Classification

OpenSUN3D Workshop at ICCV, 2025

Chunsheng (Jason) Zuo and Michael Guerzhoy

Breaking Symmetry When Training Transformers

NAACL Student Research Workshop, 2024

Sajjad Hashemi, Michael Guerzhoy, and Noah Paulson

Toward Learning Latent-Variable Representations of Microstructures by Optimizing in Spatial Statistics Space

ICLR Tiny Papers, 2024

Cameron Smith, Yiping (Sandra) Wang, and Michael Guerzhoy

Synthetic Datasets for Exploring How ConvNets and ViTs Classify Images When Colour is An Important Cue

CRV Workshops, 2024

Michael Guerzhoy and Jackson Kaunismaa

How Do ConvNets Understand Image Intensity?

ICLR Tiny Papers, 2023

ML for Understanding Creativity, Psychology, and Cognitive Science

What makes a chess move "brilliant"? What makes a Wordle game "amusing"? I use machine learning to understand human creativity, cognitive biases, and social psychology.

Ronaldo Luo, Gary Liang, Cindy Liu, Adam Kabbara, Minahil Bakhtawar, Kina Kim, and Michael Guerzhoy

Automatically Detecting Amusing Games in Wordle

ICCC 2025

Gary Liang, Adam Kabbara, Cindy Liu, Ronaldo Luo, Kina Kim, and Michael Guerzhoy

Semantic, Orthographic, and Phonological Biases in Humans' Wordle Gameplay

Findings of the ACL: IJCNLP-AACL 2025

Juho Kim and Michael Guerzhoy

Exploring the "Honour Culture" Theory Using Social Media Data

Workshop on Social Influence in Conversations at EMNLP, 2024

Sophie Lee and Michael Guerzhoy

Quantifying the Complexity of Literary Fiction

NLP for Digital Humanities Workshop at EMNLP, 2024

Kamron Zaidi and Michael Guerzhoy

Predicting User Perception of Move Brilliance in Chess

ICCC 2024 Featured in New Scientist and CBC Radio

Pavlos Constas, Vikram Rawal, Matthew Honorio Oliveira, Andreas Constas, Aditya Khan, Kaison Cheung, Najma Sultani, Carrie Chen, Micol Altomare, Michael Akzam, Jiacheng Chen, Vhea He, Lauren Altomare, Heraa Murqi, Asad Khan, Nimit Amikumar Bhanshali, Youssef Rachad, and Michael Guerzhoy

Toward A Reinforcement-Learning-Based System for Adjusting Medication to Minimize Speech Disfluency

ML for Cognitive and Mental Health Workshop at AAAI, 2024

AI and Data Science Applications

From materials design to travel modeling to scientific literature analysis, I work on applying machine learning and data science to diverse real-world problems.

Mengli (Dawn) Duan and Michael Guerzhoy

Automatically Extracting Scientific Metrics with LLMs: A Case Study of ImageNet Papers

Workshop at NeurIPS 2025

Andy Cai, Sajad Hashemi, Noah Paulson, and Michael Guerzhoy

Latent Representation of Microstructures using Variational Autoencoders with Spatial Statistics Space Loss

ECLR Workshop at ICCV, 2025

Sajjad Hashemi, Michael Guerzhoy, and Noah Paulson

Toward Learning Latent-Variable Representations of Microstructures by Optimizing in Spatial Statistics Space

ICLR Tiny Papers, 2024

Anthony Hein, May Jiang, Vydhourie Thiyageswaran, and Michael Guerzhoy

Random Forests for Opponent Hand Estimation in Gin Rummy

EAAI 2021

Michael Guerzhoy and Aaron Hertzmann

Learning Latent Factor Models of Travel Data for Travel Prediction and Analysis

Canadian Conference on AI (AI 2014) Best Paper Award

Pedagogy of Introductory Programming

How do we teach programming effectively? I develop assignments and study teaching methods for introductory computer science courses.

Michael Guerzhoy

Synergies between Intro to Data Science and Intro to Programming via Purely Functional Programming

Statistical Society of Canada (SSC) Annual Meeting, 2025

Michael Guerzhoy

Getting Used to Pointers with Pointer Drills

ITiCSE 2025

Michael Guerzhoy

A Tournament for Pong AI Engines

Nifty Assignments at SIGCSE 2018

Michael Guerzhoy, Jackie Chi Kit Cheung, and François Pitt

Automatically Solving SAT/TOEFL Synonym Questions with Computational Linguistics

Nifty Assignments at SIGCSE 2017

Pedagogy of Data Science

How does data science education relate to computer science education? I explore pathways for students to engage with data science and computational thinking.

Tabitha Belshee, Adam Chang, Nebil Ibrahim, Mikako Inaba, Nikoo Karbassi, Angelo Kayser-Browne, Hye Jee Kim, Rachel Kim, Seungjae Ryan Lee, Natalia Orlovsky, and Michael Guerzhoy

Improving Current and Future Offerings of a Data Science Course through Large-Scale Observation of Students

SIGCSE 2021

Michael Guerzhoy

Introduction to Data Science as a Pathway to Further Study in Computing

ICER 2019

Pedagogy of Machine Learning

I design model AI assignments and explore effective ways to teach machine learning concepts, emphasizing interpretation of models and thinking about data.

Michael Guerzhoy and Liam Ernst-Selway

GPT is Coming to Class: A Song

Education Program at NeurIPS 2025 (on YouTube!)

Michael Guerzhoy

Understanding How Neural Networks See (And Read): A Slide Deck

Education Program at NeurIPS 2025

Michael Guerzhoy

Occam's Razor and Bender and Koller's Octopus

Workshop on Teaching NLP at ACL, 2024

Stephen Keeley and Michael Guerzhoy

Predicting and Preventing Deaths in the ICU: Designing and Analyzing an AI System

Model AI Assignments at EAAI 2020

Michael Guerzhoy, Lisa Zhang, and Georgy Noarov

AI Education Matters: Building a Fake News Detector

AI Matters 5(3), 2019

Michael Guerzhoy and Lisa Zhang

Building a Fake News Detector

Model AI Assignments at EAAI 2019

Michael Guerzhoy and Renjie Liao

Understanding How Recurrent Neural Networks Model Text

Model AI Assignments at EAAI 2018

Michael Guerzhoy

Neural Networks for Face Recognition with TensorFlow

Model AI Assignments at EAAI 2018

Statistics

Tracy Qian, Max Piasevoli, and Michael Guerzhoy

Automatic Model Selection using Wasserstein Generative Adversarial Networks

Statistical Society of Canada (SSC) Annual Meeting, 2024

Michael Guerzhoy

"Medium-n studies" in computing education conferences

Koli Calling 2023

Screenshot from the TV show Marketplace

For an episode of CBC Marketplace, we analyzed data from food safety inspections of locations of restaurant chains nationwide, and produced rankings of restaurants, for each city, and nationwide. We show how to combine data from different cities, in which inspector standards and levels of compliance vary, by modelling the data of the number of violations detected using quasi-Poisson regression, where the city and the chain are covariates. We subsequently worked on fitting hierarchical Bayesian models to the data to identify more differences between chains and model the data better.

Episode video: Canada's Restaurant Secrets, broadcast on Apr 11, 2014 on CBC. Watch for the Poisson regression formula at 5min 48sec! (See screenshot, or watch on youtube.)

Technical report: Michael Guerzhoy and Nathan Taback, Ranking Restaurant Chains by the Number of Health Violations Found during Inspections.

Contributed conference talk: Hierarchical Bayesian Models for Uncertainty-Quantified Ranking of Restaurant Chains by Food Safety Compliance (French version), at the 43rd Annual Meeting of the Statistical Society of Canada, June 2015, Halifax, NS.

Computer Vision (pre-2020)

ConvNets for Photo Orientation Detection

Visualization of the reason a photo with birds was classified as upright

We apply a ConvNet to the task of photo orientation detection, and produce visualizations to help demonstrate how the ConvNet accomplishes the task.

Paper: Ujash Joshi and Michael Guerzhoy, Automatic Photo Orientation Detection with Convolution Neural Networks, in Proc. of the Conference on Computer and Robot Vision (CRV 2017), May 2017, Edmonton, Alberta.

Computer Vision for Speech Analysis

Rotary phone

If you compute the spectrogram of a sound signal, you can treat it like an image (kind of) and apply object detection algorithms to analyze it. Specifically, I was working on phone classification.

Project report (MSc paper): Michael Guerzhoy, Boosting Local Spectro-Temporal Features for Speech Analysis, 2010. (Online abstract.)

Background Colour Detection/Rectangular Object Detection

Several photos lying on a brown background

For the background colour detection part, we describe a way to use the fact that the background colour appears in patches and the fact that we can predict the edge statistics of the background/non-background boundary.

We also describe a perceptual organization based rectangle detection algorithm, and use a large synthetically-generated set to tune the parameters.

The intended application is streamlining of the process of scanning in documents like photos and business cards using a flatbed scanner.

Paper: Michael Guerzhoy and Hui Zhou. Segmentation of Rectangular Objects Lying on an Unknown Background in a Small Preview Scan Image. In Proc. of the Canadian Conference on Computer and Robot Vision (CRV 2008), May 2008, Windsor, Ontario.

Patent: Michael Guerzhoy and Hui Zhou. Method and apparatus for detecting objects in an image. U.S. Patent 8,098,936, issued Jan 17, 2012.
Patent: Michael Guerzhoy and Hui Zhou. Method and apparatus for detecting objects in an image. U.S. Patent 8,433,133, issued Apr 30, 2013.

Photo Orientation Detection

A photo of a city, rotated by 180 degrees. Caption on the bottom reads 'this side up'

We developed a system that determines the orientation of the input photo (from 0, 90, 180, and 270 degrees). You can try it if you have an Epson scanner.

Patent: Michael Guerzhoy and Hui Zhou. Method and system for automatically determining the orientation of a digital image. U.S. Patent 8,094,971, issued Apr 30, 2013.

A screenshot of scanner software

Youtube review: "Auto-photo orientation: I've tested this feature and it works good... Without this feature checked, you need to place the upper-left-hand corner of the photo face down in the lower-left corner of the scanner. With this feature turned on (or checked), you can place any corner of the photo, face down in the lower-left corner of the scanner, and the Epson Perfection does a good job of making sure the photo is right-side-up after scanning. This is a good feature when you can't remember which corner of the photo you need to place down."