Martin Q. Ma

Martin Q. Ma 

Martin Q. Ma
M.S. student, Carnegie Mellon University
Email: qianlim [at] cmu (dot) edu

Google Scholar
DBLP
GitHub
Twitter

Research Summary

My primary interest is deep representation learning, including 1) fair representation learning; 2) self-supervised learning; and 3) multimodal learning.

Research Topics

Primary: self-supervised learning, multimodal learning (vision, speech and text), and fairness.
Secondary: unsupervised learning, contrastive learning, domain adaptation, visual recognition, speech recognition, time-series analysis, sentiment analysis, interpretable models, variational inference, causal inference, mutual information.

About Me

I am currently a M.S. student in School of Computer Science at Carnegie Mellon University, where I am supervised by Dr. Louis-Philippe Morency and Dr. Ruslan Salakhutdinov. I also work with Dr. Kun Zhang. Previously, I studied at Brandeis University where I was supervised by Dr. Pengyu Hong.

Highlights

Papers, codes, slides, blogs, and talk videos are available here.
- ICLR 2021, NeurIPS 2020 SSL Workshop: Relative Predictive Coding, a stabler, higher performing loss objective for better self-supervised contrastive learning.
- EMNLP 2020: Multimodal Routing, a method for interprete multimodal language analysis for sentiment analysis and emotion recognition.
- ICASSP 2020, (Oral) NeurIPS 2019 SEDL Workshop: Complex Transformer, a strong performing network for complex-valued time-series analysis.