mark goldstein
PhD Candidate, Computer Science, NYU
email: goldstein [at] nyu [dot] edu
pronouns: he/him/his
About Me
I am a PhD candidate at NYU Courant Institute of Mathematical Sciences, CILVR group, advised by Rajesh Ranganath and Thomas Wies, and a student researcher at Google DeepMind. I work on deep generative models and machine learning for health and science. I’m also a part of the STAT research group at the NYU Center for Data Science. Currently I organize the “generative model foundations” weekly meeting at Courant which brings together CS and math researchers to understand recent generative models together (PIs Eric Vanden-Eijnden, Joan Bruna, and Jon Niles-Weed).
I’ve started looking for postdoc positions + jobs for summer/fall 2025. Feel free to reach out about this!
feel free to anonymously give me comments/suggestions/feedback here
On my mind
- discrete diffusion models!
News
(November 2024) giving a talk in Chi Jin’s group at Princeton. Topic: Continuous and Discrete Diffusion.
(October 2024) giving a talk in Pieter Abbeel’s group at UC Berkeley. Topic: Continuous and Discrete Diffusion.
[Contrasting with Symile: Simple Model-Agnostic Representation Learning for Unlimited Modalities]
with Adriel Saporta, Aahlad Puli, and Rajesh Ranganath
Published in NeurIPS 2024
(Summer-Fall 2024) student researcher at Google DeepMind NYC with Will Grathwohl!
Scalable Interpolant Transformers
with Nanye (Willis) Michael Albergo, Nicholas Boffi, Eric Vanden-Eijnden, and Saining Xie
Published in ECCV 2024
(Spring 2024) honored to receive the Henning Biermann Prize for teaching by a PhD student at NYU Courant!
What’s the score? Automated Denoising Score Matching for Nonlinear Diffusions
with Raghav Singhal and Rajesh Ranganath
Published in ICML 2024
Stochastic interpolants with data-dependent couplings
with Michael Albergo, Nick Boffi, Rajesh Ranganath, and Eric Vanden-Eijnden
Spotlight in ICML 2024
Probabilistic forecasting with stochastic interpolants and Föllmer processes!
with Yifan Chen, Mengjian Hua, Michael Albergo, Nicholas Boffi, and Eric Vanden-Eijnden
Published in ICML 2024
A dynamic risk score for early prediction of cardiogenic shock using machine learning
with Yuxuan Hu, Rajesh Ranganath, and many others
Published in European Heart Journal, Acute Cardiovascular Care. Also available on arxiv.
QTNet: Predicting Drug-Induced QT Prolongation with Artificial Intelligence-Enabled Electrocardiograms
with Hao Zhang, Rajesh Ranganath, and many others
Published in Journals of the American College of Cardiology, Clinical Electrophysiology
(Fall 2023) taking part @ Flatiron Institute’s upcoming second workshop on Measure Transport, Sampling, and Diffusions!
(Fall 2023) lecture on diffusions + flows @ NYU course, Inference and Representations
(Fall 2023) talk on diffusions + flows @ Decisions, Risk and Operations ML reading group at Columbia, organized by Hongseok Namkoong!
(Summer 2023) We are running the second iteration of the workshop on Spurious Correlations, Invariance, and Stability at ICML 2023!
Where to Diffuse, How to Diffuse, and How to Get Back: Automated Learning for Multivariate Diffusions
With Raghav Singhal and Rajesh Ranganath
Published in ICLR 2023
(Fall 2022) lecture on diffusions @ Yann LeCun’s deep learning course at NYU!
(Fall 2022) talk on diffusions @ the Flatiron Institute’s workshop on Sampling, Transport, and Diffusions!
(Summer 2022) excited to be co-organizing the ICML Workshop on Spurious Correlations, Invariance, and Stability!
Survival Mixture Density Networks
with Xintian Han and Rajesh Ranganath
Published in Machine Learning for Healthcare 2022
(Summer 2022) glad to continue at Apple Health AI for the summer!
Learning Invariant Representations with Missing Data
With Aahlad Puli, Adriel Saporta, Olina Chau, Jorn-Henrik Jacobsen, Rajesh Ranganath, and Andrew C. Miller
Published in CLeaR (Causal Learning and Reasoning) 2022 and appeared in NeurIPS 2021 DistShift Workshop
(Fall 2021) selected as a recipient of the NeurIPS 2021 Outstanding Reviewer Award. Glad to be a part of it!
Inverse-Weighted Survival Games
With Xintian Han, Aahlad Puli, Thomas Wies, Adler J. Perotte, and Rajesh Ranganath
Published in NeurIPS 2021
(Summer 2021) working with Apple’s Health AI team this summer supervised by Andy Miller and team!
(photo by Julian Chokkattu/Digital Trends)
Understanding Failures in Out-of-Distribution Detection with Deep Generative Models
With Lily H. Zhang and Rajesh Ranganath
Published in ICML 2021 and appeared in RobustML workshop @ ICLR 2021
(Fall 2020) I qualified! Upgrade from Student to Candidate.
(Fall 2020) after some time away from harvard cs, happy to help out Prof Nada Amin with the harvard AI/PL seminar
(Fall 2020) the deep learning course I TA’ed in spring 2020 for Yann LeCun and Alfredo Canziani is now up on Alf’s github page, check out all of Alf’s wonderful teaching materials and thanks to students for your notetaking
(visualization by Alfredo Canziani)
X-CAL: Explicit Calibration for Survival Analysis
With Xintian Han, Aahlad Puli, Adler J. Perotte, and Rajesh Ranganath
Published in NeurIPS 2020
(Summer 2019) I’m working in Emtiyaz Khan’s Approximate Bayesian Inference group at RIKEN AIP in Tokyo!
(2018) MacCracken Fellow, NYU Graduate School of Arts and Sciences, Five years of PhD funding.
mentoring
- Abhipsha Das (Master’s Thesis at NYU, 2024, on diffusions for text)
- Shraddha Jain (current Master’s student, 2024, research on VAEs)
- Nina Mortensen (Masters Thesis at NYU, 2024, research on VAEs, now at Fauna Robotics)
- Kyle Adams (Undergradate from University of Florida, 2022, research on survival analysis, now a math phd at UF)
- mentor for Women in Data Science Datathon, Cambridge, 2021.
- tutored at a few iterations of NYU AI School, meant for first and second year undergraduates in the NYC area as a first exposure to AI/ML
I usually review for
- NeurIPS, ICML, ICLR, AAAI, AISTATS, CLeaR, JMLR, misc workshops
In a previous life
I was a research assistant and teaching fellow in the computer science department at Harvard SEAS. I am still an on/off TF for the harvard undergrad ML course. Between Harvard and NYU, I worked with the CoCoSci group at MIT BCS. Previous to that, I studied music composition, improvisation, and theory at New England Conservatory with Anthony Coleman, Stratis Minakakis and Ran Blake. I am still involved with music and rehearse with Gamelan Kusuma Laras a classical Javanese ensemble that performs the repetoire of the courts of Central Java.
playlist
[sept 7, 2024] Djoko Walujo - gender - Ldr. Pangkur sl. manyura
[sept 2, 2024] Zé Kéti - Poema de Botequim