Y. Cem Subakan

Y. Cem Subakan, Assistant Professor, Université Laval, Computer Science and Software Engineering Department Affiliate Assistant Professor, Concordia University, Computer Science and Software Engineering Department Associate Academic Member, Mila

I obtained my B.Sc. and M.Sc. degrees from Bogazici University EE department, respectively in 2011 and 2013. In Bogazici, I worked with A. Taylan Cemgil and Bulent Sankur. I got my PhD from Computer Science Department of University of Illinois at Urbana-Champaign. My PhD advisor was Paris Smaragdis. In Mila, I did a postdoc with Laurent Charlin. I am one of the core contributors to the Speechbrain project. I am co-organizing ICASSP 2024 workshop XAI-SA, and I will be the general chair of IEEE Workshop of Machine Learning for Signal Processing 2025. You can see my google scholar here.

Students / Collaborators

• Francesco Paissan

• Eleonora Mancini

• Firat Oncel

• Shubham Gupta

• Maab Elrashid

• Luca Della Libera

• Pooneh Mousavi

• Jacob Comeau

• Mathis Rezzouk

• Anthony Deschenes

We currently do not have open calls. But shoot me an email if you have experience in speech and audio, or methodological machine learning research.

Research Interests

• Time series modeling

• Interpretable Machine Learning, Explainability, Posthoc Interpretations

• Audio Source separation / Speech Enhancement

• Deep Generative Models

• Machine Learning, Statistical Learning

• Bayesian inference

• Non-convex optimization

• Method of Moments based learning algorithms for Latent Variable Models

Representative papers

• Listenable Maps for Audio Classifiers

• REAL-M: Towards Speech Separation on Real Mixtures

• Attention is All You Need in Speech Separation

• Learning the Base Distribution in Implicit Generative Models

• Generative Adversarial Source Separation

• Spectral Learning of Mixture of HMMs

Teaching

• Machine Learning for Signal Processing (Fall 2024)

• Machine Learning for Signal Processing (Fall 2023)

Notes from my timeline

• Our paper is accepted to NeurIPS 2024!

• Our paper is accepted to EMNLP 2024!

• Three of our papers are accepted to Interspeech 2024!

• Our paper Listenable Maps for Audio Classifiers is accepted to ICML 2024 as an oral presentation.

• We organized XAI-SA 2024, an IEEE ICASSP 2024 sattelite workshop.

• Our paper REAL-M: Towards Speech Separation on Real Mixtures is accepted to ICASSP 2022.

• Our paper Attention is All You Need in Speech Separation is accepted to ICASSP 2021.

• Our paper ‘On the Effectiveness of Two-Step Learning for Latent-Variable Models‘, is accepted to MLSP 2020.

• Our paper is accepted to MICCAI 2020.

• Our paper won the best student paper award in MLSP 2017.

• I have visited Columbia in Fall 2015.

• I have done an internship at Analog Devices/Lyric Labs in Summer 2015.

• I am visiting Daniel Hsu at Columbia U. in Spring 2015.

• I am a qualified PhD student! (passed qual) Here's my qual presentation.

• Our paper ‘‘Spectral Learning of Mixture of Hidden Markov Models’’ is accepted to NIPS 2014!

• Our DSP paper got accepted!

• We have presented a paper in NIPS2013 spectral learning workshop.

• I have successfully defended my M.Sc. thesis.

Contact

email: