For those of wondering if there will be a tea-talk this Friday I have good news - the answer is yes! Our next speakers will be Rémi Leblond and Jean-Baptiste Alayrac, both PhD students at INRIA, working with Simon Lacoste-Julien. Please come to AA6214 on July 7 at 13:45!
1) Searnn, Rémi LeblondTitle: SeaRNN: Training RNNs with Global-Local LossesAbstract: We propose SEARNN, a novel training algorithm for recurrent neural networks (RNNs) inspired by the “learning to search” (L2S) approach to structured prediction. RNNs have been widely successful in structured prediction applications such as machine translation or parsing, and are commonly trained using maximum likelihood estimation (MLE). Unfortunately, this training loss is not always an appropriate surrogate for the test error: by only maximizing the ground truth probability, it fails to exploit the wealth of information offered by structured losses. Further, it introduces discrepancies between training and predicting (such as exposure bias) that may hurt test performance. Instead, SEARNN leverages test-alike search space exploration to introduce global-local losses that are closer to the test error. We demonstrate improved performance over MLE on three different tasks: OCR, spelling correction and text chunking. Finally, we propose a subsampling strategy to enable SEARNN to scale to large vocabulary sizes. Bio: Rémi is a second year Ph.D. student at INRIA in the SIERRA team under the supervision of Simon Lacoste-Julien. He's interested in very large scale optimization for new complex machine learning models. His current focus is on parallel and distributed algorithms and structured prediction using neural networks.After graduating from Ecole Polytechnique and Corps des Mines (French MPA), he worked in the Big Data field for three years, as a Data Scientist at SpotRight in the US (2011-2012) and as a Data Engineer at the Ministry of Defense (2013-2015).
2) Joint discovery of objects and manipulation actions, JB Alayrac
Title: Joint discovery of objects and manipulation actions
Abstract:
Many human activities involve object manipulations aiming to modify the object state. Examples of common state changes include full/empty bottle, open/closed door, and attached/detached car wheel. In this work, we seek to automatically discover the states of objects and the associated manipulation actions. Given a set of videos for a particular task, we propose a joint model that learns to identify object states and to localize state-modifying actions. Our model is formulated as a discriminative clustering cost with constraints. We assume a consistent temporal order for the changes in object states and manipulation actions, and introduce new optimization techniques to learn model parameters without additional supervision. We demonstrate successful discovery of seven manipulation actions and corresponding object states on a new dataset of videos depicting real-life object manipulations. We show that our joint formulation results in an improvement of object state discovery by action recognition and vice versa.
Bio:
Jean-Baptiste Alayrac received the MS degree in computer science in Ecole Normale Superieure (ENS), in Paris in 2014. He is currently working toward the PhD degree in the research teams WILLOW and SIERRA at INRIA Paris under the supervision of Josef Sivic, Ivan Laptev and Simon Lacoste-Julien. His research focuses on structured prediction from vision and natural
language.