New England Machine Learning Day

Seventy researchers, students and professionals recently participated in the New England Machine Learning Hackathon: Hacking Bias in ML. Students traveled predominantly from nearby universities — Harvard, MIT, Northeastern, Boston University, Boston College — but also from as far as the University of Virginia, Carnegie Mellon and Stanford.  

Team leaders defined five areas of bias and discrimination in machine learning to address during the Hackathon.

Accent Discrimination led by Jay Liu, Microsoft
Pre-Trial Fairness, led Sam Corbett-Davies, Stanford
Word Biases led by Max Leiserson and Elena Jakubiak, Microsoft Research
Visual Biases in Border Patrol Stops led by Genevieve Patterson, Microsoft Research
Equity in Higher Education and the Future of Work, led by Sergio Marrero, Caila

In six hours, the teams formed, defined a user, aligned on a solution, built storyboards, and in some cases, developed websites. There was buzz, laughter, and hard work, which may have benefitted from the nearby espresso bar. The winning team, Pre-Trial Fairness, took home Xbox FIFA bundles. The team built a “Challenge the Bias” website to “decipher the biases in current algorithms that help decide bail, sentences and parole of a defendant.” The judges appreciated the clear explanation of the types of data used and ways to increase fairness across gender and race in analyzing the data. Congratulations Sam Corbett-Davies (Stanford), Yaovi Ayeh (Dell EMC), Danielle Dean (Microsoft), Frances Ding (Harvard), Yunxin Fan (Harvard), Anshuman Pandey (CMU), Aditthya Ramakrishnan (Next Tech Lab), Harini Suresh (MIT), Marina S. (community) and Lorenzo Vitale (BU)!

The Word Bias team created a Hacking Bias in Word Choice website. The Accent Team pitched and is building an open repository for under-represented accents which limits the capabilities of voice recognition everywhere. The Visual Bias team highlighted ways to help border patrol agents decrease unconscious biases against those wearing non-normative, less frequently seen outfits. The Equity in Education team brainstormed ways to advance individuals with less traditional educational degrees and training and connecting to jobs they can do today, tomorrow with a bit of work, and in the future with more serious preparation and study.

One grad student after returning to campus wrote, “Thank you for … such an amazing hackathon on a really interesting topic in ML. It was totally worth coming all the way from Pittsburgh to Boston and attend this hackathon.” (It took him and his colleague 18 hours by bus!)

See our Hacking Bias in Machine Learning recap video (2m)  above with comments from our event mentors and judges. Our team mentors were Dr. Adam Kalai (Microsoft Research); Dr. Lester Mackey (Microsoft Research); and one of the judges, Elaine Harris (Hacking Discrimination MIT Alumni organizer and President, Breakthrough Marketing Technology).  In addition to Elaine Harris, our esteemed judging panel included: Dr. Sasha Constanza-Chock, Associate Professor of Civic Media at MIT; Dr. Stefanie Jegelka, X-Consortium Career Development Assistant Professor at MIT EECS; Jamie MacLennan, Microsoft, Partner Director, Azure Machine Learning; Dr. D. Sculley, Google, Engineering Manager, Machine Learning Team.

The sixth annual New England Machine Learning Day took place on the following day, May 12, 2017.  The event brought together more than 300 local machine learning researchers from over a dozen universities and research institutes. Eight talks were given by notable local academics on a variety of machine learning problems ranging from neural networks to computer vision to social networks.  Thirty-six students presented posters during a lively poster session at lunch.  

The organizing committee comprised:  David Cox (Harvard); Adam Tauman Kalai (Microsoft Research); Ankur Moitra (MIT); and Kate Saenko (Boston University).  The Poster Chairs were Mike Hughes, Harvard University and Genevieve Patterson, Microsoft Research.

One Northeastern graduate student who attended both said, “The hackathon was a great experience for me and I enjoyed every second of it. The New England ML day was also very inspiring. If for any similar event you need a volunteer, please let me know. I’d be more than happy to help and be a part of these great events.”

Stay tuned for next year; planning has begun for our seventh New England Machine Learning Day and second ML Hackathon!