Travis Dick

GHC 6008

tdick@cs.cmu.edu

About Me

I am a fifth year PhD student at Carnegie Mellon University in the Computer Science Department where I am fortunate to be advised by Nina Balcan. Before that, I completed my MSc at the University of Alberta, advised by Richard Sutton and András György.

I am interested in building on the theory and practice of Machine Learning to accommodate several modern requirements of learning systems. In particular, I have projects on new requirements stemming from three distinct sources: making the best use of available data, applying learning tools to problems beyond standard prediction, and incorporating social values. I have also worked on actor-critic methods for Reinforcement Learning, Online Learning, and distributed learning.

Here is my cv.

Manuscripts

Maria-Florina Balcan, Travis Dick, Ritesh Noothigattu, Ariel D. Procaccia. Envy-free Classification.

Conference Papers

Maria-Florina Balcan, Travis Dick, Colin White. Data-Driven Clustering via Parameterized Lloyd's Families. NIPS 2018.

Maria-Florina Balcan, Travis Dick, Ellen Vitercik. Dispersion for Data-Driven Algorithm Design, Online Learning, and Private Optimization. FOCS 2018.

Maria-Florina Balcan, Travis Dick, Tuomas Sandholm, Ellen Vitercik. Learning to Branch. ICML 2018.

Maria-Florina Balcan, Travis Dick, Yingyu Liang, Wenlong Mou, Hongyang Zhang. Differentially Private Clustering in High-Dimensional Euclidean Spaces. ICML 2017.

Travis Dick, Mu Li, Venkata Krishna Pillutla, Colin White, Maria-Florina Balcan, Alex Smola. Data Driven Resource Allocation for Distributed Learning. AIStats 2017.

Maria-Florina Balcan, Travis Dick, Yishay Mansour. Label Efficient Learning by Exploiting Multi-class Output Codes. AAAI 2017.

Travis Dick, András György, Csaba Szepesvári. Online Learning in Markov Decision Processes with Changing Cost Sequences. ICML 2014.

Roshan Shariff, Travis Dick. Lunar Lander: A Continuous-Action Case Study for Policy Gradient Actor Critic Algorithms. RLDM 2013 (poster).

Patrick Pilarski, Travis Dick, Richard Sutton. Real-time Prediction Learning for the Simultaneous Actuation of Multiple Prosthetic Joints. ICORR 2013.

Travis Dick, Camilo Perez, Azad Shademan, Martin Jagersand. Realtime Registration-based Tracking via Approximate Nearest Neighbour Search. RSS 2013.

PhD Thesis

I proposed my PhD Thesis in December 2019 titled "Machine Learning: Social Values, Data Efficiency, and Beyond Prediction". I am fortunate to have Maria-Florina Balcan, Yishay Mansour, Tom Mitchell, and Ariel Procaccia as my thesis committee.
Related links:

Workshop Papers

Kareem Amin, Travis Dick, Alex Kulesza, Andres Medina, Sergei Vassilvitskii. Private Covariance Estimation via Iterative Eigenvector Sampling. NIPS Privacy Preserving Machine Learning Workshop 2018.

Maria-Florina Balcan, Travis Dick, Ritesh Noothigattu, Ariel D. Procaccia. Envy-free Classification. NIPS Workshop on Ethical, Social, and Governance Issues in AI 2018.

Maria-Florina Balcan, Travis Dick, Ellen Vitercik. Dispersion for Private Optimization of Piecewise Lipschitz Functions. ICML Privacy in Machine Learning and AI Workshop 2018.

Maria-Florina Balcan, Travis Dick, Ellen Vitercik. Differentially Private Algorithm Configuration. ICML Private and Secure Machine Learning Workshop 2017.

Travis Dick, Mu Li, Venkata Krishna Pillutla, Colin White, Maria-Florina Balcan, Alex Smola. Data Driven Resource Allocation for Distributed Learning. AAAI Distributed Machine Learning Workshop 2017.

Maria-Florina Balcan, Travis Dick, Yishay Mansour. On the Geometry of Output-code Multi-class Learning. ICML Data Efficient Machine Learning Workshop 2016.

Talks

Invited Talks
Dispersion for Data-Driven Algorithm Design, Online Learning, and Private Optimization.
  • Talk at the Civil Engineering Machine Learning Seminar at CMU (2019).
  • Graduation Day Talk at Information Theory and Applications Workshop (2019).
  • China Theory Week (2018).
Envy-free Classification
  • Poster at Google Fairness in Machine Learning Workshop (2018).
Label Efficient Learning by Exploiting Multi-class Output Codes
  • Talk at Picky Learners Workshop at ICML (2017). (slides)
Other Talks
Machine Learning: Social Values, Data Efficiency, and Beyond Prediction
  • Talk at University of Pennsylvania CIS Department (2019).
  • Talk at Toyota Technical Institute at Chicago (2019). (slides)
Private Covariance Estimation via Iterative Eigenvector Sampling
  • Poster at NeurIPS Privacy Preserving Machine Learning Workshop (2018).
Envy-free Classification
  • Poster at NeurIPS Workshop on Ethical, Social, and Governance Issues in AI (2018).
Data-driven Clustering via Parameterized Lloyd’s Families
  • Spotlight Talk at NeurIPS (2018).
Dispersion for Data-Driven Algorithm Design, Online Learning, and Private Optimization
  • Talk at Microsoft Research NYC (2018). (slides)
  • Talk at Google Research NYC (2018).
  • Talk at FOCS (2018). (slides)
Learning to Branch
  • Poster presentation at ICML (2018).
Differentially Private Algorithm Configuration
  • Poster Presentation at Privacy in Machine Learning and AI Workshop at ICML (2018).
  • Poster Presentation at Private and Secure Machine Learning Workshop at ICML (2017).
Data Driven Resource Allocation for Distributed Learning
  • AI Lunch Seminar at CMU (2017).
  • Talk at AAAI Distributed Machine Learning Workshop (2017). (slides)
Label Efficient Learning by Exploiting Multi-class Output Codes
  • Poster Presentation at AAAI (2017). (poster)
  • Talk at ICML Data Efficient Machine Learning Workshop (2016). (short video)
  • Theory Lunch Seminar at CMU (2016). (video)
Online Learning in Markov Decision Processes with Changing Cost Sequences
  • Talk at ICML (2014).
  • Poster Presentation at RLDM (2013).

Teaching

In the Spring semesters of 2015 and 2016, I was a TA for Introduction to Machine Learning (10-601). In 2015 the course was taught by Nina Balcan and Tom Mitchell, and in 2016 it was taught by Nina Balcan and William Cohen. I received a TA Award from the Machine Learning Department for TAing 10-601 in Spring 2016.