Tamara Broderick: Variational Bayes and Beyond: Bayesian Inference for Big Data (ICML 2018 tutorial)
Abstract:
Bayesian methods exhibit a number of desirable properties for modern data analysis---including (1) coherent quantification of uncertainty, (2) a modular modeling framework able to capture complex phenomena, (3) the ability to incorporate prior information from an expert source, and (4) interpretability. In practice, though, Bayesian inference necessitates approximation of a high-dimensional integral, and some traditional algorithms for this purpose can be slow---notably at data scales of current interest. The tutorial will cover modern tools for fast, approximate Bayesian inference at scale. One increasingly popular framework is provided by "variational Bayes" (VB), which formulates Bayesian inference as an optimization problem. We will examine key benefits and pitfalls of using VB in practice, with a focus on the widespread "mean-field variational Bayes" (MFVB) subtype. We will highlight properties that anyone working with VB, from the data analyst to the theoretician, should be aware of. In addition to VB, we will cover recent data summarization techniques for scalable Bayesian inference that come equipped with finite-data theoretical guarantees on quality. We will motivate our exploration throughout with practical data analysis examples and point to a number of open problems in the field.
Presented by Tamara Broderick (MIT)
Webpage: http://www.tamarabroderick.com/tutorial2018icml.html
Видео Tamara Broderick: Variational Bayes and Beyond: Bayesian Inference for Big Data (ICML 2018 tutorial) канала Steven Van Vaerenbergh
Bayesian methods exhibit a number of desirable properties for modern data analysis---including (1) coherent quantification of uncertainty, (2) a modular modeling framework able to capture complex phenomena, (3) the ability to incorporate prior information from an expert source, and (4) interpretability. In practice, though, Bayesian inference necessitates approximation of a high-dimensional integral, and some traditional algorithms for this purpose can be slow---notably at data scales of current interest. The tutorial will cover modern tools for fast, approximate Bayesian inference at scale. One increasingly popular framework is provided by "variational Bayes" (VB), which formulates Bayesian inference as an optimization problem. We will examine key benefits and pitfalls of using VB in practice, with a focus on the widespread "mean-field variational Bayes" (MFVB) subtype. We will highlight properties that anyone working with VB, from the data analyst to the theoretician, should be aware of. In addition to VB, we will cover recent data summarization techniques for scalable Bayesian inference that come equipped with finite-data theoretical guarantees on quality. We will motivate our exploration throughout with practical data analysis examples and point to a number of open problems in the field.
Presented by Tamara Broderick (MIT)
Webpage: http://www.tamarabroderick.com/tutorial2018icml.html
Видео Tamara Broderick: Variational Bayes and Beyond: Bayesian Inference for Big Data (ICML 2018 tutorial) канала Steven Van Vaerenbergh
Показать
Комментарии отсутствуют
Информация о видео
4 сентября 2018 г. 16:59:15
02:17:51
Другие видео канала
MLSS 2019 David Blei: Variational Inference: Foundations and Innovations (Part 1)Variational AutoencodersInterview with Professor Tamara Broderick, MITICLR Debate with Leslie Kaelbling (ICLR 2019)Probabilistic Programming and Bayesian Modeling with PyMC3 - Christopher FonnesbeckA visual guide to Bayesian thinkingRobert Geirhos: ImageNet-trained CNNs are biased towards texture (ICLR 2019 talk)Ian Goodfellow: Adversarial Machine Learning (ICLR 2019 invited talk)J. Z. Kolter and A. Madry: Adversarial Robustness - Theory and Practice (NeurIPS 2018 Tutorial)Tutorial Session: Variational Bayes and Beyond: Bayesian Inference for Big DataNonparametric Bayesian Methods: Models, Algorithms, and Applications IFast Quantification of Uncertainty and Robustness with Variational BayesBayesian Nonparametrics Part I - Tamara Broderick - MLSS 2015 TübingenProf. David Blei - Probabilistic Topic Models and User BehaviorNorthern Frontier: In conversation with Prof. Tamara BroderickCalculus of Variations ft. Flammable MathsChristopher D Manning: A Neural Network Model That Can Reason (ICLR 2018 invited talk)Bernhard Schölkopf: Learning Causal Mechanisms (ICLR invited talk)Variational Inference: Foundations and Modern Methods (NIPS 2016 tutorial)Andrew Gelman - Solve All Your Statistics Problems Using P-Values