ML-NYC

Speaker Series and Happy Hour

Next Meeting: 11/15, 4:00pm @ The Flatiron Institute:

162 5th Ave (entrance on 21st between 5th and 6th)

Speaker: Samory Kpotufe

Columbia University

Title: Adaptivity in Domain Adaptation and Friends.

Abstract: Domain adaptation, transfer, multitask, meta, few-shots, representation, or lifelong learning … these are all important recent directions in ML that all touch at the core of what we might mean by ‘AI’. As these directions all concern learning in heterogeneous and ever-changing environments, they all share a central question: what information a data distribution may have about another, critically, in the context of a given estimation problem, e.g., classification, regression, bandits, etc. Our understanding of these problems is still rather fledgeling. We plan to present both some recent positive results and also some negative ones. On one hand, recent measures of discrepancy between distributions, fine-tuned to given estimation problems (classification, bandits, etc) offer a more optimistic picture than existing probability metrics (e.g. Wasserstein, TV) or divergences (KL, Renyi, etc) in terms of achievable rates. On the other hand, when considering seemingly simple extensions to choices between multiple datasets (as in multitask), or multiple prediction models (as in Structural Risk Minimization), it turns out that minimax oracle rates are not always adaptively achievable, i.e., using just the available data without side information. These negative results suggest that domain adaptation is more structured in practice than captured by common invariants considered in the literature.

Bio: Samory Kpotufe is an associate professor in Statistics at Columbia University. He graduated (Sept 2010) in Computer Science at the University of California, San Diego, advised by Sanjoy Dasgupta. He then was a researcher at the Max Planck Institute for Intelligent Systems. At the MPI he worked in the department of Bernhard Schoelkopf, in the learning theory group of Ulrike von Luxburg. Following this, he spent a couple years as an Assistant Research Professor at the Toyota Technological Institute at Chicago. He then spent 4 years at ORFE, Princeton University as an Assistant Professor. Recently he was a visiting member at the Institute of Advanced Study from January to July 2020.

The Flatiron Institute Guest Protocols

  • By entering our buildings the person implicitly attests that they do not have symptoms consistent with COVID and they are not knowingly COVID positive.

  • Masks should be worn in indoor group settings such as meetings, conferences, and workshops. Masks should also be worn in restrooms.

  • Walk with a government issued ID. Guests will need to show their ID when they arrive at the building.

Happy Hour

5:00-6:00pm, following the talk

Location:
The Flatiron Institute

About ML-NYC

The Machine Learning New York City (ML-NYC) Speaker Series and Happy Hour is a monthly event for machine learning practitioners, researchers, and students to meet and watch talks from leading researchers in the field. Each event will feature a New York City-based speaker presenting their work, followed by a happy hour at a nearby bar. The ML-NYC Speaker Series is open to anyone interested in machine learning, and we encourage everyone to attend, whether you are a beginner or an expert in the field.

ML-NYC is generously supported by the Flatiron Institute, the NYU Center for Data Science and the Columbia Data Science Institute.

Organizers

ML-NYC is organized by professors David Blei and Joan Bruna along with PhD students David Brandfonbrener, Claudia Shi, Min Jae Song, and Keyon Vafa.