| Session 1 [video] | |
| 08:50 - 09:00 | Opening remarks |
| 09:00 - 09:35 | Tatiana Lopez-Guevara |
| 09:35 - 09:40 | Spotlight: Object Abstraction in Visual Model-Based Reinforcement Learning |
| 09:40 - 09:45 | Spotlight: Unsupervised Neural Segmentation and Clustering for Unit Discovery in Sequential Data |
| 09:45 - 09:50 | Spotlight: Incorporating Domain Knowledge About XRF Spectra into Neural Networks |
| 09:50 - 10:30 | Break |
| Session 2 [video] | |
| 10:30 - 11:05 | Josh Tenenbaum |
| 11:05 - 11:40 | Sanja Fidler |
| 11:40 - 11:50 | Spotlight: CvxNets: Learnable Convex Decomposition |
| 11:50 - 12:00 | Spotlight: Spatially Invariant, Label-free Object Tracking |
| 12:00 - 13:30 | Lunch |
| Session 3 [video] | |
| 13:30 - 14:05 | Niloy Mitra |
| 14:05 - 14:40 | Danilo Rezende |
| 14:40 - 15:30 | Posters |
| 15:30 - 16:15 | Coffee break |
| Session 4 [video] | |
| 16:15 - 17:15 | Panel discussion |
| 17:15 - 17:30 | Closing remarks |
Invited talks
Seeing in a World of Objects, Physics, and Agents:
Generative models for human perception and human-like machine perception
Josh Tenenbaum - Professor at the Department of Brain and Cognitive Sciences, MIT (USA).
Inverse Physics for Robotic Manipulation of Liquids
Tatiana Lopez-Guevara (Edinburgh Centre for Robotics -UK)
Our brains are able to exploit coarse physical models of fluids to quickly adapt and solve everyday manipulation tasks. However, developing such capability in robots, so that they can autonomously manipulate fluids adapting to different conditions remains a challenge.
In this talk, I will present how a Robot can use an internal model based on simulation to infer properties of real liquids, via direct and indirect interactions, using Bayesian Optimization. I will introduce the challenges of using approximate simulation as a forward model, and the usefulness of the estimations to perform a pouring task on different containers and liquid properties.
Bio Tatiana Lopez-Guevara is a 3rd year PhD student in Robotics and Autonomous Systems at the Edinburgh Centre for Robotics, UK. Her interests are in the application of intuitive physics models for robotic reasoning and manipulation of deformable objects.
NeuralCG: Exploring Generative 3D Modeling for Content Creation
Niloy Mitra (University College London - UK)
Creating high-quality is expensive as it requires significant manual input to generate diverse yet plausible geometric and topological variations, with and without textures. Hence, there is a strong demand for generative models producing novel, diverse, and realistic 3D shapes along with associated part semantics and structure. A key challenge towards this goal is how to accommodate diverse shape, including both continuous deformations of parts as well as structural or discrete alterations which add to, remove from, or modify the shape constituents and compositional structure. In this talk I will present StructureNet, a hierarchical graph network which can directly encode shapes represented as such n-ary graphs; can be robustly trained on large and complex shape families; and be used to generate a great diversity of realistic structured shape geometries. The learned latent spaces enable several structure-aware geometry processing applications, including shape generation and interpolation, shape editing, or shape structure discovery directly from un-annotated images, point clouds, or partial scans. For more details, please visit this page.
Bio Niloy Mitra received his Master’s and Ph.D. in Electrical Engineering from Stanford University under the guidance of Leonidas Guibas. His research interest lies in representing, reconstructing, analyzing, editing and fabricating 3D shapes for computer graphics and vision applications. He is known for his work in local/global registration, symmetry and regularity detection, structure-aware geometry processing, and generative 3D modeling. His current research focuses on developing machine learning frameworks for generating high-quality geometric and appearance models for CG applications. Niloy received the 2019 Eurographics Outstanding Technical Contributions Award, the 2015 British Computer Society Roger Needham Award, and the 2013 ACM Siggraph Significant New Researcher Award. He leads the Smart Geometry Group at University College London and also Adobe Research London. For more detail, please visit this page.
Non-Supervised Learning and Decision Making
Danilo J. Rezende - Research Scientist, DeepMind (UK).
A.I. Data Factory for A.I.
Sanja Fidler - Assistant Professor at the University of Toronto and Director of AI at NVIDIA (Canada).