Michael Guerzhoy

Mike

Lecturer, Department of Computer Science, University of Toronto
Office: BA5244. Current office hours: by appointment.
Email: guerzhoy at cs.toronto.edu

In addition to teaching, I work in machine learning, computer vision, and applied statistics and take on data science consulting projects.

Teaching (current): CSC411 — Machine Learning and Data Mining.
Teaching (recent): CSC180, Fall 2014/2015/2016, STA303/STA1002, Summer 2016, C4M: Computing for Medicine, 2016, CSC321, Winter 2016 (won the CSSU award for excellence in teaching), CSC320, Winter 2015, CSC165, Summer 2014.

Project courses supervised: Twitter Hashtag Recommendation and Analysis (Ramaneek Gill, CSC494/495, 2015-2016); RNN and Spectral Feature Based Music Analysis and Generation (Karo Castro-Wunsch, CSC492, 2016); Photo Orientation Detection with ConvNets (Ujash Joshi, CSC494/CSC495, 2016); Demonstration in Reinforcement Learning (Omobola Okesanjo, CSC494, Fall 2016); Computer Vision for Camera Trap Data (Joshua Samson-Seltzer, GGR417, 2016-2017)
Current research: current collaborative projects with students; hierarchical Bayesian models for ranking; regularization of log-linear models.

Graduate student supervision: Thi Hai Van Do (MScAC, University of Toronto).

My assignments around the web: SAT Synonyms, presented at Nifty Assignments at SIGCSE (2017). Fun with RNNs (CSC321, 2016), used at the University of Toronto (2017) and Hacettepe University (2017). Transfer Learning with Neural Networks (CSC321/CSC411, 2016-2017), used in part at Udacity as part of the Self-Driving Car Engineer Nanodegree (2016-).

AlexNet implementation+weights in TensorFlow

The UofT Data Science Team; a tournament for Pong AIs, 2016 (the 2015 tournament).

Recent media mentions: quoted in MIT Technology Review (Jun. 2017) on teaching machine learning with TensorFlow and on the TensorFlow ecosystem; the story also appeared in Business Insider (Jul. 2017); quoted in The Varsity (Feb. 2017) on AI for literature search and on careers in machine learning and data science; quoted in The Cannon on "curving" and grading policies; Computing for Medicine profiled in UofT News (Mar. 2016).

Back in grad school, I used to co-ordinate the weekly CSGSBS cookie breaks.

My Erdős-Bacon number is (with some latitude allowed) at most 8.

Derandomizing Bogosort: A Very Serious Webpage.





Selected Projects: Research · Industrial R&D · Consulting

       marketplace  

Statistical models for ranking

For an episode of CBC Markeplace, 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 to 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.

   
catcartoon

Mike Twohy, The New Yorker, June 5, 1995



       to paris  

Latent factor models of human travel

We decompose the likelihood of traveling from A to B into three factors: the desirability of B as a destination, the affinity between source A and destination B, and the individual-varying propensity to travel the distance between A and B. By analyzing a large dataset of geotagged Flickr photos, we estimate the desirabilities of destinations on the map and affinities between locations, as well as discover clusters of individuals with varying propensities to travel large distances. We analyze the learned affinity factors to discover travel patterns within and across linguistic boundaries. We also analyze a dataset of Shanghai taxi trips.

Paper: Michael Guerzhoy and Aaron Hertzmann, Learning Latent Factor Models of Human Travel. At NIPS Workshop on Social Network and Social Media Analysis: Methods, Models and Applications (Social 2012), Dec. 2012, Lake Tahoe, Nevada.

Paper: Michael Guerzhoy and Aaron Hertzmann, Learning Latent Factor Models of Travel Data for Travel Prediction and Analysis. In Proc. of the Canadian Conference on Aritifical Intelligence (AI 2014), May 2014, Montreal, Quebec. Best Paper Award.

Project website

   


       photoOriBirds  

ConvNets for Photo Orientation Detection: improving performance and visualizing the system

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.

   


       phone  

Computer vision for speech analysis

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.)

   



       catdetect  

Object detection

I've worked on several object detection projects. I've been particularly interested in image features. (On the left is the output of a cat detector I made).

   



       segphotos  

Background colour detection/rectangular object detection

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.

   



       arbatover  

Photo orientation detection

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.

driver

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.

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."

   

Teaching

CSC411 — Machine Learning and Data Mining, Winter 2017 (lectures). Course website.

GGR417 — Honours Thesis (Geography/Environmental Science), 2016-2017 (co-supervising Joshua Samson-Seltzer)

CSC494 — Computer Science Project, Fall 2016 (supervising Omobola Okesanjo)

CSC495 — Computer Science Project, Fall 2016 (supervising Ujash Joshi)

CSC180 — Introduction to Computer Programming, Fall 2016 (lectures). Course website.

CSC494 — Computer Science Project, Summer 2016 (supervising Ujash Joshi) official course description.

STA303/STA1002 — Methods of Data Analysis II, Summer 2016 (lectures). Course website, official course description.

C4M — Computing for Medicine, Winter 2016 - Summer 2016 (design and delivery of the workshops, with Michelle Craig). Course website, description.

CSC495 — Computer Science Project, Winter 2016 (supervising Ramaneek Gill) official course description.

CSC492 — Computer Science Implementation Project, Winter 2016 (co-supervising Karo Castro-Wunsch) official course description.

CSC321 — Introduction to Neural Networks and Machine Learning, Winter 2016 (lectures). Course website, official course description.

CSC494 — Computer Science Project, Fall 2015 (supervising Ramaneek Gill) official course description.

CSC180 — Introduction to Computer Programming, Fall 2015 (lectures). Course website, official course description.

CSC320 — Introduction to Visual Computing, Winter 2015 (lectures). Course website, official course description.

CSC180 — Introduction to Computer Programming, Fall 2014 (lectures). Course website, official course description.

CSC165 — Mathematical Expression and Reasoning for Computer Science, Summer 2014 (lectures). Course website, official course description.

CSC180 — Introduction to Computer Programming, Fall 2010 (lectures). Here are some ideas for short presentations about advanced topics.

CSC180 — Introduction to Computer Programming, Fall 2009 (lectures). The labs from that offering are online.

CSC180 — Introduction to Computer Programming, Fall 2008 (head TA). Tutorials




"Guerzhoy" is pronounced guer-ZHOY. "Zh" denotes the voiced postalveolar fricative, i.e., the "s" in "measure."

The illustration "Amberley Phone" by Flickr user dotsandspaces licenced under CC.

Valid HTML 4.01 Transitional