Seminario Aprendizaje de Máquinas

On the unreasonable effectiveness of the Sinkhorn algorithm

Event Date: Jan 14, 2020 in Seminario Aprendizaje de Máquinas, Seminars

Abstract: This talk concerns Sinkhorn algorithm, broadly understood as the iterative scaling of a matrix that realizes the solution of an entropy regularized linear program subjected to row and column constraints. I will present new theoretical and applied results that demonstrate the effectiveness of this procedure in two contexts: first, Sinkhorn algorithm implements the solution of an entropy regularized version of optimal transport. I will show this regularization substantially improves sample complexity over the unregularized case, a result that helps explain the popularity of this...

Read More

An Introduction to Reinforcement Learning and Reward Machines.

Event Date: Jan 07, 2020 in Seminario Aprendizaje de Máquinas, Seminars

In Reinforcement Learning (RL), an agent is guided by the rewards it receives from the reward function. Unfortunately, it may take many interactions with the environment to learn from sparse rewards, and it can be challenging to specify reward functions that reflect complex reward-worthy behavior. We propose using reward machines (RMs), which are automata-based representations that expose reward function structure, as a normal form representation for reward functions. We show how specifications of reward in various formal languages, including LTL and other regular languages, can be...

Read More

Latent distance estimation for random geometric graphs.

Event Date: Jan 07, 2020 in Seminario Aprendizaje de Máquinas, Seminars

Abstract: Random geometric graphs are a popular choice for a latent points generative model for networks. Their definition is based on a sample of $n$ points $X_1,X_2,\cdots,X_n$ on the Euclidean sphere~$\mathbb{S}^{d-1}$ which represents the latent positions of nodes of the network. The connection probabilities between the nodes are determined by an unknown function (referred to as the “link” function) evaluated at the distance between the latent points. We introduce a spectral estimator of the pairwise distance between latent points and we prove that its rate of convergence is...

Read More

Seminario Conjunto: Aprendizaje de Máquinas y Modelamiento Estocástico. Mini Workshop: Computational Optimal Transport

Event Date: Nov 30, 1999 in Seminario Aprendizaje de Máquinas, Seminars, Stochastic Modeling

Lugar: Sala (B04, piso -1, Beauchef 851) Fecha: Lunes 26, noviembre, 2018. Hora: 1500 – 1700hrs Presentadora: Dr. Elsa Cazelles Título: Statistical properties of barycenters in the Wasserstein space Hora: 1500hrs Abstract: In this work, we discuss the analysis of data in the form of probability measures on R^d. The aim is to provide a better understanding of the usual statistical tools on this space endowed with the Wasserstein distance. The first order statistical analysis is a natural notion to consider, consisting of the study of the Fréchet mean (or barycenter). In particular, we...

Read More

“Privacy Preserving Machine Learning”

Event Date: Nov 30, 1999 in Seminario Aprendizaje de Máquinas, Seminars

Abstract: Los diversos avances en cloud computing y aprendizaje automático han permitido difundir el uso de modelos cliente-servidor, donde un cliente proporciona sus datos y un servidor tiene un modelo de aprendizaje automático para hacer alguna inferencia sobre los datos del cliente. Sin embargo, dado distintos aspectos de la privacidad de los datos, el cliente puede no querer revelar la información requerida por el servidor o incluso dejar que el servidor conozca el resultado del modelo. Esta es una situación común cuando el tipo de información involucrada corresponde a datos médicos,...

Read More

Optimal transport and applications to Data Science

Event Date: Jan 17, 2018 in Seminario Aprendizaje de Máquinas, Seminars

Abstract: Optimal transport (OT) provides rich representations of the discrepancy between probability measures supported on geometric spaces. Recently, thanks to the development of computational techniques, OT has been used to address problems involving massive datasets, as an alternative to usual KL-divergence based approaches. In this talk I will introduce the OT problem and comment on its elementary duality properties. Then, I will present the entropy regularized problem and its (fast) solution via Sinkhorn iterations. Finally, I will overview two applications to Data Science: first,...

Read More