State representation learning for robotics control
Dr. Natalia Diaz, Robotics and Computer Vision, ENSTA ParisTech
Abstract: Representation learning algorithms are designed to learn abstract features that characterize data. State representation learning (SRL) focuses on a particular kind of representation learning where learned features are in low dimension, evolve through time, and are influenced by actions of an agent. In other words, SRL algorithms are built to capture factors of variation influenced by an agent in a given environment and project it into a disentangled and low dimensional space. As the representation learned captures the variation in the environment generated by agents, this kind of representation is particularly suitable for robotics and control scenarios. Moreover, the low dimension helps to overcome the curse of dimensionality and provides easier interpretation and utilization for both humans and other algorithms. Therefore, SRL can improve both human-machine interaction, performance and speed in policy learning algorithms such as reinforcement learning. I will present an overview covering the state-of-the-art on state representation learning in the most recent years considering methods that involve interaction with the environment and their applications in robotics control tasks. I will also present a robustness benchmark on Baxter robot on unsupervised state representation learning with robotic priors .
Natalia Díaz Rodríguez, PhD is Computer Engineer by the University of Granada (UGR, Spain) and has a Double PhD from Abo Akademi (Finland) (together with UGR) on Artificial Intelligence and Semantic and Fuzzy Modelling for Human Behaviour Recognition in Smart Spaces. She has worked on R&D at CERN (Switzerland), Philips Research (Netherlands) at the Personal Health Dept., done a postdoctoral stay at University of California Santa Cruz, and worked in Silicon Valley at Stitch Fix (San Francisco, CA) -a recommendation service for personalized fashion delivery with humans in the machine learning loop).
She has participated in a range of international projects such as being Management Committee member of EU COST (European Cooperation in Science and Technology) Action AAPELE.EU (Algorithms, Architectures and Platforms for Enhanced Living Environments, www.aapele.eu), was Google Anita Borg Scholar 2014, Heidelberg Laureate Forum 2014 & 2017 fellow, and obtained the Nokia Scholarship. Currently she is researcher and lecturer at ENSTA ParisTech at the Robotics and Computer Vision group , and is working on Deep and Reinforcement Learning for visual state representation learning for the robotics DREAM project .
 Unsupervised state representation learning with robotic priors: a robustness benchmark Timothée Lesort, Mathieu Seurin, Xinrui Li, Natalia Díaz Rodríguez, David Filliat. ENSTA ParisTech, France. ArXiv: https://arxiv.org/abs/1709.05185 DEMO: https://www.youtube.com/watch?v=wFd0yJuJlQ4
Location: JRC Sevilla.
REGISTRATION REQUIRED FOR EXTERNAL VISITORS BEFORE April 13th. PLEASE REGISTER HERE.