Inference and Learning via Randomly Perturbed Optimization
Venue: Via Sommarive 5 - Polo Ferrari 1 (Povo, TN) - Room A210
- Stefano Ermon, Stanford University
Probabilistic inference is a central problem in machine learning and computer science. To date, only a handful of distinct methods have been developed, most notably (MCMC) sampling, decomposition, and variational methods. In this talk, I will discuss recent advances in new approaches based on solving randomly perturbed optimization problems, which can provide more favorable tradeoffs between accuracy and cost. This also leads to a new class of generative models where samples are produced by following randomly perturbed gradients of the data distribution, which can be estimated with score matching. Our framework allows flexible model architectures, requires no sampling during training or the use of adversarial methods, and provides a learning objective that can be used for principled model comparisons. Our models produce samples comparable to GANs on MNIST, CelebA and CIFAR-10 datasets, achieving a new state-of-the-art inception score of 8.91 on CIFAR-10.
About the Speaker
Stefano Ermon is an Assistant Professor of Computer Science in the CS Department at Stanford University, where he is affiliated with the Artificial Intelligence Laboratory, and a fellow of the Woods Institute for the Environment. His research is centered on techniques for probabilistic modeling of data, inference, and optimization, and is motivated by a range of applications, in particular ones in the emerging field of computational sustainability. He has won several awards, including four Best Paper Awards (AAAI, UAI and CP), a NSF Career Award, ONR and AFOSR Young Investigator Awards, a Sony Faculty Innovation Award, an AWS Machine Learning Award, a Hellman Faculty Fellowship, Microsoft Research Fellowship, and the IJCAI Computers and Thought Award. Stefano earned his Ph.D. in Computer Science at Cornell University in 2015.
Note: Stefano Ermon will be at DISI on november 27 and be available to meet collegues at DISI. Please email Roberto Sebastiani for organize meetings.
Contact: roberto.sebastiani [at] unitn.it (Roberto Sebastiani)