Selected Publications & Patents

(2025). Score-of-Mixture Training: One-Step Generative Model Training Made Simple via Score Estimation of Mixture Distributions. Spotlight Proceedings of the 42nd International Conference on Machine Learning (ICML 2025).

arXiv Link Openreview Link Code

(2025). Score Estimation for Generative Modeling. Massachusetts Institute of Technology.

Thesis

(2024). Lifted Residual Score Estimation. 2024 ICML SPIGM Workshop.

Paper

(2024). Model-Code Separation Architecture for Data Compression Using Sum-Product Networks. US Patent App (US20260017501A1).

Link

(2023). Score-based Source Separation with Applications to Digital Communication Signals. 37th Conference on Neural Information Processing Systems (NeurIPS 2023).

arXiv Link Openreview Link Project Webpage Code

(2023). Self-Supervised Representations for Singing Voice Conversion. 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

Paper Audio Samples

(2022). System and Method for Detecting Adversarial Attacks. US Patent App.

Link

(2022). Architecture for Variable Bitrate Neural Speech Codec with Configurable Computation Complexity. 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

Paper

(2022). Image Compression using Sum-Product Networks. Massachusetts Institute of Technology.

Thesis

(2020). Detecting Audio Attacks on ASR Systems with Dropout Uncertainty. INTERSPEECH 2020.

arXiv Link Conference Link

(2019). LAP-based Video Frame Interpolation. 2019 IEEE International Conference on Image Processing (ICIP).

Paper