CV for Michael C. Hughes

PDF | source | Last updated: March 13 2024

Research Expertise

  • 2012-

    Machine Learning

    • Unsupervised learning, Semi-supervised learning, Supervised learning
    • Probabilistic models: Bayesian nonparametrics, topic models, hidden Markov models, deep generative models
    • Posterior estimation methods: variational methods, Markov chain Monte Carlo

  • 2017-

    Clinical Informatics

    • Tasks: Phenotype discovery, personalized outcome prediction, automated assisted diagnosis
    • Data Types: Time series of vitals and lab results from EHR, diagnosis and procedure codes, sociodemographics, imaging (esp. echocardiograms)
    • Applications: intensive care early warning systems, suggesting treatments for depression, heart disease diagnosis

  • 2012-

    Human Activity Analysis

    • Sensor time-series, motion capture, video, images

Education

  • Brown University

    2016

    Ph.D., Computer Science.

  • Brown University

    2012

    M.S., Computer Science.

  • Olin College of Engineering

    2010

    B.S. Electrical & Computer Engineering

Research Experience

  • Assistant Professor of Computer Science

    2018 - present

    Tufts University, Medford, MA

  • Postdoctoral fellow: Machine learning to improve clinical decisions in healthcare

    2016 - 2018

    Adviser: Prof. Finale Doshi-Velez (Harvard)

    • Developed semi-supervised models for characterizing and treating depression (with Dr. Perlis and Dr. McCoy).
    • Applied time-series models to predict ventilator interventions in the ICU for public dataset of >36,000 patients.
    • Created methods for training deep models so they are more interpretable to clinicians or other users.

  • Postdoc project: Estimating carbon biomass from LiDAR waveforms

    2016

    Adviser: Prof. Erik Sudderth & Prof. Jim Kellner (Brown U., Ecology & Evolutionary Biology)

  • Ph.D. thesis: Reliable and scalable variational inference for Bayesian nonparametrics

    2016

    Adviser: Prof. Erik Sudderth

    • Thesis Title: Reliable and scalable variational inference for nonparametric mixtures, topics, and sequences
    • Developed optimization algorithms for Bayesian nonparametric models that scale to millions of examples.
    • Optimized lower bound on marginal likelihood, thus penalizing too simple and too complex explanations.
    • Escaped local optima via data-driven proposals that add useful new clusters and remove redundant ones.
    • Applied to topic models of 2 million NY Times articles and sequential models of the whole human genome.
    • Implemented algorithms in open-source package: Bayesian Nonparametrics for Python (BNPy).

  • Master's project: Sequential Models for Video and Motion Capture

    2012

    Adviser: Prof. Erik Sudderth

    • Developed methods to discover common actions from many videos of humans performing household exercises.
    • Improved existing inference algorithms with data-driven Metropolis-Hastings proposals.

Highlighted Publications

Superscripts indicate mentored student's status: u = undergraduate, m = masters, d = Ph.D. student, b = post-bacc, c = medical student. Complete publication list at end of this document.

Highlighted Preprints

Honors and Awards for Research

Honors and Awards for Peer Review

  • 2023

    Top 10 Percent Reviewer Award, AISTATS 2023

    • Recognized as one of top 10 percent of all 2500 expert reviewers at a top international machine learning conference.

  • 2022

    Highlighted Reviewer Award, ICLR 2022

    • Recognized as a top reviewer at ICLR, a top-tier international machine learning conference.

  • 2022

    Top 10 Percent Reviewer Award, AISTATS 2022

    • Recognized as one of top 10 percent of all 2500 expert reviewers at a top-tier international machine learning conference.

  • 2020

    Top 10 Percent Reviewer Award, NeurIPS 2020

    • Recognized as one of top 10 percent of more than 3500 expert reviewers at the top international machine learning conference.

  • 2019

    Top 400 Reviewer Award, NeurIPS 2019

    • Recognized as one of top 400 of more than 3500 expert reviewers at the top international machine learning conference.

  • 2018

    Top 200 Reviewer Award, NeurIPS 2018

    • Recognized as one of top 200 of more than 3500 expert reviewers at the top international machine learning conference.

Current Funding Support

  • C2D2AI: Pilot Investigation for Neuroimaging Biomarkers

    9/1/23 - 1/31/25

    Alzheimer's Drug Discovery Foundation (ADDF)

    • Full title: Covert Cerebrovascular Disease Detected by Artificial Intelligence (C2D2AI): Pilot Investigation for Pragmatic Neuroimaging Biomarkers for Future Stroke and Dementia Risk
    • Co-Investigators: David Kent (PI, Tufts Medical) and Wansu Chen (Kaiser Permanente Southern Cal.)
    • Total Amount: \$599,788

  • Machine Learning Models for Human Performance Prediction

    9/1/20 - present

    U.S. Army NSRDEC, Natick, MA (via Tufts CABCS)

    • Full title: Statistical and Machine Learning Models for Data Reduction and Human Performance Prediction
    • Co-Investigators: Eric Miller (co-PI, Tufts ECE) and Shuchin Aeron (co-PI, Tufts ECE)
    • Total Amount: roughly \$496,098 per year
    • 1 year contract renewed each year for 4 years in a row
    • Part of larger cross-institution MASTR-E project funded by US Army

Past Funding Support

  • Autonomous Cognitive Technologies for Novelty in Open Worlds

    11/15/19 - 5/15/23

    DARPA SAIL-ON Program

  • Amortized Inference for Large-Scale Graphical Models

    9/1/19 - 8/31/22

    NSF CISE: Robust Intelligence: Small

    • Co-Investigators: Liping Liu (PI, Tufts CS) and Thomas Stopka (Tufts Public Health)
    • Total Amount: \$399,923

  • A Benchmark De-identified Echocardiogram Database

    5/1/21 - 4/30/22

    Pilot Grant from Tufts CTSI

    • Full title: A Benchmark De-identified Echocardiogram Database for Studying Automated Diagnoses
    • Co-Investigators: Benjamin Wessler (PI)
    • Tufts CTSI = Clinical and Translational Science Institute

  • Estimating the societal value of COVID-19 therapeutics

    1/1/21 - 12/31/21

    Tufts Medical Center (Originating Sponsor: Pfizer)

    • Co-Investigators: Peter Neumann (Tufts CEVR) and Joshua Cohen (Tufts CEVR)

  • The value of predictive analytics during the COVID epidemic

    7/1/20 - 6/30/21

    Tufts Springboard Award (Tufts Univ. Provost's Office)

    • Full title: Demonstrating the value of a proposed Tufts-led predictive analytics and comparative effectiveness research network during the COVID epidemic
    • Co-Investigators: David Kent (Tufts Medical) and Jessica Paulus (Tufts Medical)
    • Total Amount: \$50,000

  • Estimating Individual Treatment Effects

    7/1/19 - 6/30/20

    Tufts Collaborates Award (Internal)

    • Title: 'Estimating Individual Treatment Effects from Randomized Clinical Trials using Machine Learning'
    • Co-Investigators: David Kent (Tufts Medical Center)
    • Total Amount: \$50,251

Invited Talks

Professional Service

  • Senior Area Chair

    • 2024 - CHIL
    • 2023 - CHIL

  • Area Chair

    • 2023 - MLHC and ML4H
    • 2022 - CHIL and MLHC
    • 2021 - CHIL and MLHC

  • Senior Program Committee / Meta-Reviewer

    • 2021 - AAAI
    • 2020 - AAAI

  • Program Committee / Reviewer

    • 2023 - NeurIPS, NeurIPS Datasets&Benchmarks, AISTATS, ICLR
    • 2022 - NeurIPS, NeurIPS Datasets&Benchmarks, AISTATS, ICLR
    • 2020 - NeurIPS, AISTATS, ICLR, MLHC
    • 2019 - NeurIPS (reviewer award), AISTATS, ICLR
    • 2018 - NeurIPS (reviewer award), AAAI, AISTATS, ICLR, AMIA CRI
    • 2017 - NeurIPS, ICML, AAAI
    • 2016 - NeurIPS
    • 2015 - NeurIPS, ICML
    • 2014 - NeurIPS, ICML
    • 2013 - NeurIPS (reviewer award)

  • 2018

    Workshop Organizer: ML4H at NeurIPS 2018

    • Machine Learning for Health workshop at NeurIPS '18 (NeurIPS ML4H 2018).
    • Full-day workshop with invited keynotes, accepted papers/posters, and lively panel discussions.
    • Provided a forum for interdisciplinary interaction between clinicians, statisticians, and computer scientists.
    • Helped with website, PR, and continuity in peer-review process from previous years.

  • 2018

    Workshop Organizer: BNP at NeurIPS 2018

  • 2017

    Workshop Organizer: ML4H at NeurIPS 2017

    • Machine Learning for Health workshop at NIPS '17 (NIPS ML4H 2017)
    • Full-day workshop with invited keynotes and panels involving clinicians, statisticians, and computer scientists.
    • Organized peer-review process for 118 submitted papers.

  • 2016

    Workshop Organizer: BNP at NeurIPS 2016

    • Practical Bayesian Nonparametrics workshop at NIPS '16.
    • Full-day workshop with invited speakers, contributed talks, two panel discussions, and lively poster session.
    • Led decisions on >25 submitted papers based on peer review.

Mentorship

  • Postdoctoral Mentees

    Tufts University

    • Michael T. Wojnowicz, from 2020 - 2023
    • - Tufts DISC Data Scientist, primary adviser M. C. Hughes
    • - Papers: ICML '22
    • - Next position: Research associate at Harvard Univ., Dept. of Biostatistics, advised by Jeff Miller

  • Doctoral Mentees

    Tufts University

    • Zhe Huang, CS, from 2019 - present (spring 2024 expected)
    • - Topic: Overcoming Limited Labeled Data in the Wild
    • - Awards: Outstanding Academic Scholarship Award from Tufts School of Engineering (one per year)
    • - Papers: AISTATS '23 (oral), MLHC '23, MLHC '21, NeurIPS D&B '21
    • Preetish Rath, CS, from 2019 - present (fall 2024 expected)
    • - Topic: Addressing False Alarms and Missingness in Clinical Prediction Models
    • - Papers: AISTATS '22, IMLH workshop '23
    • Patrick Feeney, CS, from 2020 - spring 2025 (expected)
    • - Topic: Self-supervised and Supervised Contrastive Learning in Open Worlds
    • - Papers: TMLR '23, arXiv '23
    • Kyle Heuton, CS, from 2020 - fall 2025 (expected)
    • - Topic: Training spatiotemporal models to learn where to intervene
    • - Papers: IMLH workshop '23
    • Ethan Harvey, CS, from 2023 - fall 2026 (expected)
    • - Topic: Time-to-event prediction of dementia from neuroimages

  • Co-advised Doctoral Research Mentees

    Tufts University

    • Kevin Cheng, ECE, from 2019 - fall 2022
    • - Primary adviser: Eric Miller (Tufts ECE)
    • - Topic: Optimal Transport methods for Time Series modeling
    • - Papers: NeurIPS '21, IEEE Trans Signal Proc. '23, ICASSP '20
    • - Next position: Takeda, Principal AI/ML Research Scientist

  • Masters Program Research Mentees

    Tufts University

    • Yu Liu, MS DS with thesis, from 2020 - 2021
    • - Topic: An Evaluation Pipeline for Heterogeneous Treatment Effect Prediction
    • - Next Position: Merck
    • Xi Chen, MS CS with project, from 2020 - 2021
    • - Topic: Bayesian Nonparametric Mixture Models for Missing Data
    • - Next Position: CS PhD student at Rutgers

  • Post-bacc Research Mentees

    Tufts University

    • Ally Lee, BS CS from Tufts, in 2020
    • - Topic: Bayesian Analysis of Autoregressive Models for Multi-Site Hospital Admission Forecasting
    • - Next Position: Software Engineer at Hubspot, Boston, MA
    • Lily H. Zhang, BS from Harvard, from 2019 - 2020
    • - Topic: 'Any Parameter Encoders for Topic Models: Variational Encoders that amortize across models as well as data'
    • - Next Position: PhD candidate in CS at NYU, advised by Rajesh Ranganath

  • Undergraduate Research Mentees

    Tufts University

    • Mary-Joy Sidhom, BS CS with honors thesis, in 2022
    • - Topic: Deep Learning for Doppler Echocardiography from Limited Labeled Data
    • - Papers: AISTATS '23 (oral)
    • - Next Position: Software Engineer II at ASML
    • Hezekiah Branch, BS CS at Tufts, in 2020
    • - Tufts LSAMP fellow
    • - Topic: Supervised Learning for Clinical Multivariate Time Series

  • Harvard University SEAS

    2016-2017

    Research Mentor

    • Mentored undergraduate senior thesis projects on Bayesian nonparametric inference.
    • Frederick Widjaja. 2017 honors thesis: {Streaming Variational Inference for the Indian Buffet Process}.
    • Madhu Vijay. 2017 honors thesis: {Characterizing Posterior Uncertainty for the Indian Buffet Process}.

  • Brown University

    2014-2016

    Research Mentor

Teaching

  • Tufts CS Dept.

    Fall 2023

    Course: CS 135 Intro to Machine Learning

    • Taught core principles of machine learning to 101 students
    • Course format: 3 open-ended projects, 5 homeworks (conceptual and code questions), 2 exams

  • Tufts CS Dept.

    Spring 2023

    Course: CS 136 Statistical Pattern Recognition

    • Taught advanced statistical learning course to 36 students
    • Course format: 5 math-intensive homeworks, 5 coding-intensive homeworks, 5 quizzes

  • Tufts CS Dept.

    Fall 2022

    Course: CS 152 Bayesian Deep Learning

    • Taught special topics course to 26 students
    • Course format: weekly homeworks for first month, then 2-month open-ended team project

  • Tufts CS Dept.

    Spring 2021

    Course: COMP 136 Statistical Pattern Recognition

    • Taught advanced statistical learning course to 29 students
    • Course format: 5 math-intensive homeworks, 5 coding-intensive homeworks, 5 short quizzes, 2 exams

  • Tufts CS Dept.

    Fall 2020

    Course: COMP 135 Intro to Machine Learning

    • Taught core principles of machine learning to 95 students
    • Course format: 3 open-ended projects, 5 homeworks (conceptual and code questions), and 5 quizzes

  • Tufts CS Dept.

    Spring 2020

    Course: COMP 136 Statistical Pattern Recognition

    • Taught advanced statistical learning course to 35 students
    • Course format: 5 math-intensive homeworks, 5 coding-intensive homeworks, 5 short quizzes, 2 exams

  • Tufts CS Dept.

    Fall 2019

    Course: COMP 150 Bayesian Deep Learning

    • Taught advanced topics seminar to 23 students
    • Course format: weekly homeworks for first month, then 2-month open-ended team project

  • Tufts CS Dept.

    Spring 2019

    Course: COMP 135 Introduction to Machine Learning

    • Taught core principles of machine learning to about 50 students
    • Course format: 3 open-ended projects, weekly homeworks, and 2 exams

  • Tufts CS Dept.

    Fall 2018

    Course: COMP 150 Bayesian Deep Learning

    • Taught advanced topics seminar to about 18 students
    • Course format: weekly homeworks for first month, then 2-month open-ended team project
    • One project resulted in publication at IEEE conference (ICDL-EpiRob 2019)

  • Fall 2013

    Lead Graduate TA for CS 142: Intro to Machine Learning

    • Led weekly 1 hour recitation session to review key concepts for 50+ students.
    • Designed homework assignments and exam questions.

Outreach Experience

  • Tufts DIAMONDS Program , Medford, MA

    2021, 2022, 2023

    Research Mentor

    • Mentored two students / summer in data science research projects.

  • TEALS and Boston Latin Academy , Roxbury, MA

    2014-2016

    Volunteer AP Computer Science Instructor

    • Taught 1-2 classes / week for 2 years via TEALS "CS in every high school" initiative sponsored by Microsoft.
    • Developed hands-on lessons to excite students from diverse backgrounds about computational thinking.
    • Mentored full-time teacher Ingrid Roche as she transitioned from media arts to AP computer science (Java).

  • Harvard Humanitarian Initiative , Cambridge, MA

    2014

    Signal Program Fellow

    • Developed prototype detector for common housing structures in sub-Saharan Africa from satellite images.
    • Intended for humanitarian oversight of conflict areas where burning structures is common attack pattern.
    • Featured in TEDx talk: https://youtu.be/u7l9rBwOnwU

  • Olin College Engineering Discovery , Needham, MA

    2007-2010

    Co-Founder and Curriculum Director

    • Managed 15 undergrads in developing hands-on lessons for 4th-8th graders.
    • Hosted workshops for 30 children to design, build, and launch bottle rockets.
    • Pioneered green energy workshop which earned over \$750 in outside funding.

Industry Experience

  • Google , Mountain View, CA

    Summer 2013

    Software Engineering Intern

    • Improved walking/biking/running classifier using smartphone accelerometer data.
    • Led collection of dataset from dozens of individuals for classifier evaluation via custom Android app.

All Conference Publications (in reverse chronological order)

All Journal Publications (in reverse chronological order)

All Workshop Papers (in reverse chronological order)