Minerva logo small mpi Minerva logo small eth

Max Planck ETH Center for Learning Systems Inauguration

30th November 2015, Tübingen, Germany


Learning systems are part of our everyday life, either as software in internet search as well as in image recognition or as physical systems such as robot vacuum cleaners and autonomous vehicles. The Max Planck ETH Center for Learning Systems promotes leading experts as well as junior scientists in the interdisciplinary, pioneering research field of learning systems. It establishes a centre of excellence offering a unique platform for scientific networking and research into novel technologies at the science and technology hub in Baden-Wuerttemberg and Switzerland.

Registration upon invitation is requested by 10th November 2015 at registration@is.mpg.de.



Schedule

Morning Session Inauguration ceremony

Description From To
Welcome 11:00 12:30
Prof. Dr. Martin Stratmann
President of the Max-Planck-Society
Prof. Dr. Lino Guzzella
President of ETH Zurich
Words of Greeting
Christine Schraner Burgener
Ambassador of Switzerland in the Federal Republic of Germany
Theresia Bauer
Minister of Science, Research and the Arts Baden-Württemberg
Scientific Introduction
Prof. Dr.Thomas Hofmann
ETH Zurich, Co-Director Max Planck ETH Center for Learning Systems
Prof. Dr. Bernhard Schöplkopf
MPI for Intelligent Systems, Co-Director Max Planck ETH Center for Learning Systems
Prof. Dr. Bradley Nelson
ETH Zurich, Institute for Robotics und Intelligent Systems

Afternoon Session Inauguration Symposium

Description From To
Keynote Lecture: "On the Computational Complexity of Deep Learning" 14:00 15:00
Prof. Shai Shalev-Shwartz

School of Computer Science and Engineering, The Hebrew University, Jerusalem & VP Technology Mobileye

In recent years, deep learning based systems have led to breakthroughs in computer vision, speech recognition, and other hard AI tasks. In contrast, from the theoretical point of view, by and large, we do not understand why deep learning is at all possible, since most state of the art theoretical results show that deep learning is computationally hard. Bridging this gap is one of the most important open problems in learning theory. I will present new positive and negative results and will also discuss the need for a "practically relevant theory".

Lab Tours: Get an insight in the Departments of the MPI for Intelligent Systems 15:00 16:00
Keynote lecture: "Perceiving Neural Networks" 16:00 17:00
Prof. Matthias Bethge

Werner Reichardt Centre for Integrative Neuroscience, University of Tübingen, Institute for Theoreti-cal Physics, Bernstein Center for Computational Neuroscience, & Max Planck Institute for Biological Cy-bernetics

Let’s compete—benchmarking models in neuroscience: Computational modeling has become increasingly popular in neuroscience but it often lacks a common strategy for model comparison. Following the benchmarking approach ubiquitous in machine learning I will present three problems in neuroscience for which model comparison plays an important role: (1) Predicting where people look, (2) predicting when neurons spike, and (3) generative modeling of natural images. I will conclude with a discussion on the growing importance of Machine Learning in neuroscience and how the increasing proficiency of artificial neural networks in solving perceptual tasks opens exciting new opportunities for interaction between the two fields.