Speaker Information

Senior Director, NVIDIA AI Technology Center

Abstract

Recently statistical machine learning which is a subset of Artificial Intelligence has made great progress. The accuracy of computer vision, automatic speech recognition, image processing and language translation using machine learning and deep learning have already surpass human capability. However these technologies required huge amount of data to train. The size of the neural network has grown a few order in term of size. In order to achieve higher level artificial intelligence, we need to develop algorithms that can be trained with much smaller data size or simulated model. Also simulation models are getting more complex and thus computational hungry. Moore’s law will not able to meet the computational need of the new simulation requirement. In this talk, the author discuss about the state of art AI algorithm, application and how simulation is being used for Artificial General Intelligence and Artificial Intelligence can be used to advance and reduce the computational requirement of large scale simulation model.

Speaker Information

Senior Researcher at National Institute of Advanced Industrial Science and Technology, Japan

Abstract

Recent progress in autonomous mobile robots found that mobile robots are ready to start mobility services in real complex environments. This keynote introduces our autonomous mobile robot project for human living spaces. There are many applications of mobility services in cities; autonomous driving, food delivery robots, cleaning or security robots, etc. Essential functions of the robots' movement are ready to use in an actual situation. In this talk, we would like to discuss the missing technical part to make autonomous robots familiar in daily living spaces.

Speaker Information

Professor at NTHU and Principal Researcher at Microsoft

Abstract

Multi-modality benefits from multiple data representations provided from different perspectives. Each modality can provide a description of the object or the scene based on its unique characteristic representation. Deep learning technology has proved to be very successful for many different tasks with single-modal inputs by training models with extremely large and representative datasets. Deep multimodal learning has attracted increasing attention since it can benefit model learning from data of different modalities to boost the accuracy and robustness of the deep neural network system. In this talk, I will present real examples of applying multimodal learning to the training of face recognition and face anti-spoofing systems, including those used in Windows Hello.

Marijn van Knippenberg (Eindhoven University of Technology),
Mike Holenderski (Einhoven University of Technology),
Vlado Menkovski (Eindhoven University of Technology)

Abstract

Multi-Agent Path Finding (MAPF) is a routing problem in which multiple agents need to each find a lowest-cost collection of routes in a graph that avoids collisions between agents. This problem occurs frequently in the domain of logistics, for example in the routing of trains in shunting yards, airplanes at airports, and picking robots in automated warehouses. A solution is presented for the MAPF problem in which agents operate on an arbitrary directed graph, rather than the commonly assumed grid world, which extends support to use cases where the environment cannot be easily modeled in a grid shape. Furthermore, constraints are introduced on the start and end times of the routing tasks, which is vital in MAPF problems that are part of larger logistics systems. A Reinforcement Learning-based (RL) approach is proposed to learn a local routing policy for an agent in a manner that relieves the need for manually designing heuristics. It relies on a Graph Convolutional Network to handle arbitrary graphs. Both single-agent and multi-agent RL approaches are presented, showing how a multi-agent setup can reduce training time by exploiting the similarities in agent properties and local graph topologies.

Reference link