23/04/2021, 14:00 — 15:00 Europe/Lisbon —
Jan Peters, Technische Universitaet Darmstadt
Robot Learning - Quo Vadis?
Autonomous robots that can assist humans in situations of daily life have been a long standing vision of robotics, artificial intelligence, and cognitive sciences. A first step towards this goal is to create robots that can learn tasks triggered by environmental context or higher level instruction. However, learning techniques have yet to live up to this promise as only few methods manage to scale to high-dimensional manipulator or humanoid robots. In this talk, we investigate a general framework suitable for learning motor skills in robotics which is based on the principles behind many analytical robotics approaches. It involves generating a representation of motor skills by parameterized motor primitive policies acting as building blocks of movement generation, and a learned task module that transforms these movements into motor commands. We discuss learning on three different levels of abstraction, i.e., learning for accurate control is needed to execute, learning of motor primitives is needed to acquire simple movements, and learning of the task-dependent „hyperparameters“ of these motor primitives allows learning complex tasks. We discuss task-appropriate learning approaches for imitation learning, model learning and reinforcement learning for robots with many degrees of freedom. Empirical evaluations on a several robot systems illustrate the effectiveness and applicability to learning control on an anthropomorphic robot arm. These robot motor skills range from toy examples (e.g., paddling a ball, ball-in-a-cup, juggling) to playing robot table tennis against a human being and manipulation of various objects.
28/04/2021, 18:00 — 19:00 Europe/Lisbon —
Mikhail Belkin, Halicioğlu Data Science Institute, University of California San Diego
07/05/2021, 14:00 — 15:00 Europe/Lisbon —
Rebecca Willett, University of Chicago
Machine Learning and Inverse Problems: Deeper and More Robust
Many challenging image processing tasks can be described by an ill-posed linear inverse problem: deblurring, deconvolution, inpainting, compressed sensing, and superresolution all lie in this framework. Recent advances in machine learning and image processing have illustrated that it is often possible to learn a regularizer from training data that can outperform more traditional approaches by large margins. In this talk, I will describe the central prevailing themes of this emerging area and a taxonomy that can be used to categorize different problems and reconstruction methods. We will also explore mechanisms for model adaptation; that is, given a network trained to solve an initial inverse problem with a known forward model, we propose novel procedures that adapt the network to a perturbed forward model, even without full knowledge of the perturbation. Finally, I will describe a new class of approaches based on "infinite-depth networks" that can yield up to a 4dB PSNR improvement in reconstruction accuracy above state-of-the-art alternatives and where the computational budget can be selected at test time to optimize context-dependent trade-offs between accuracy and computation.
21/05/2021, 14:00 — 15:00 Europe/Lisbon —
Kyriakos Vamvoudakis, Georgia Institute of Technology
28/05/2021, 14:00 — 15:00 Europe/Lisbon —
Gustau Camps-Valls, Universitat de València
Physics Aware Machine Learning for the Earth Sciences
Most problems in Earth sciences aim to do inferences about the system, where accurate predictions are just a tiny part of the whole problem. Inferences mean understanding variables relations, deriving models that are physically interpretable, that are simple parsimonious, and mathematically tractable. Machine learning models alone are excellent approximators, but very often do not respect the most elementary laws of physics, like mass or energy conservation, so consistency and confidence are compromised. I will review the main challenges ahead in the field, and introduce several ways to live in the Physics and machine learning interplay that allows us (1) to encode differential equations from data, (2) constrain data-driven models with physics-priors and dependence constraints, (3) improve parameterizations, (4) emulate physical models, and (5) blend data-driven and process-based models. This is a collective long-term AI agenda towards developing and applying algorithms capable of discovering knowledge in the Earth system.