ASA Connect

 View Only
  • 1.  SLDS May webinar

    Posted 05-20-2024 19:23

    Dear Colleagues, 

    The SLDS May webinar features Dr. Andrej Risteski from Carnegie Mellon University.  Dr. Risteski will discuss about score-based losses, which are computationally appealing alternatives to maximum likelihood.  Hope to see you there!  

    Title:                         The statistical cost of score-based losses

    Speakers:                Dr. Andrej Risteski, Machine Learning Department, Carnegie Mellon University

    Date and Time:       May 30, 2024, 1:00 to 2:30 pm Eastern Time

    Registration Link:   ASA SLDS Webinar Registration Link [] 

    Abstract:                Score-based losses have emerged as a computationally appealing alternative to maximum likelihood for fitting (probabilistic) generative models with an intractable likelihood (for example, energy-based models and diffusion models). What is gained by foregoing maximum likelihood is a tractable gradient-based training algorithm. What is lost is less clear: in particular, since maximum likelihood is asymptotically optimal in terms of statistical efficiency, how suboptimal are score-based losses?

    I will survey a recently developing connection relating the statistical efficiency of broad families of generalized score losses, to the algorithmic efficiency of a natural inference-time algorithm: namely, the mixing time of a suitable Markov chain using the score that can be used to draw samples from the model. This connection allows us to understand when score-based losses come at a small statistical cost, as well as to elucidate the design space for score losses with good statistical behavior.

    Presenter:               Andrej Risteski is an Assistant Professor at the Machine Learning Department in Carnegie Mellon University. Prior to that, he was a Norbert Wiener Research Fellow jointly in the Applied Math department and IDSS at MIT. He received his PhD in the Computer Science Department at Princeton University.

    His research interests lie in the intersection of machine learning, statistics, and theoretical computer science, spanning topics like (probabilistic) generative models, algorithmic tools for learning and inference, representation and self-supervised learning, out-of-distribution generalization and applications of neural approaches to natural language processing and scientific domains. More broadly, the goal of his research is principled and mathematical understanding of statistical and algorithmic problems arising in modern machine learning paradigms.
    He is the recipient of an NSF CAREER Award, an Amazon Research Award and a Google Research Scholar Award.

    Zhihua Su, PhD
    Associate Professor
    Department of Statistics
    University of Florida