Topic:

Learning at Scale: From Manufacturing with Intelligence to Combating Global Pandemics

 

Speaker:

陳佩君 博士

Dr. Trista Chen

英業達人工智慧中心首席科學家

Chief Scientist of Machine Learning, Inventec Corp.

 

Language :

Chinese

 

Date:

2020/4/24

 

Time:

14:20~

 

Location: 

Delta building R216, NTHU

 

Abstract:

When it comes to deep learning (DL), three D’s come to mind instantly: “Big Computing, Big Data, Big Results.” The ever-more complex DL models require massive computation power and lots of training data with laborious preparation processes. Finally, big rewards are expected as a result of DL. Regarding Big Computing, as the old saying goes, "those long divided shall be united; those long united shall be divided: such is the way of the universe". Distributed Learning, including the new Federated Learning, has gained a lot of popularity with promises to deliver more without compromising each computing unit’s privacy. However, there posts significant challenges when each computing unit only has access to a small amount of labeled data. Such challenges arise at manufacturing facilities with data integrity and security considerations. Likewise, when combating global pandemics, ways to derive the best public-health result given a small amount of individual data is also challenging. Regarding Big Data, both few-shot learning and reinforcement learning show significant progress in recent years. In this talk, we share our real-world experience working with first-tier electronics manufacturing facilities on adopting a new DL framework to learn and tune deep-learning models at scale, to reduce big data requirements, to touch upon some open problems, and hopefully to reap big rewards for the future.

Short bio: 

Trista is a tech executive, entrepreneur, and artificial intelligence (AI) scientist. She is currently the Chief Scientist of Machine Learning at Inventec Corp., a world-leading computer and consumer-electronics manufacturer. She is in charge of Inventec’s AI efforts in smart manufacturing, smart health and medical AI, etc. Previously, Trista held leadership positions at startups and research labs. While leading core algorithms and data science at Cognitive Networks, which was later acquired by Vizio, she lifted the company's customer counts by more than 10x. At Intel, she conducted research on computer vision algorithm and hardware co-design and was part of the OpenCV development team. At Nvidia, she architected Nvidia's first video processor.

Trista received her Ph.D. from Carnegie Mellon University. She co-authored 26 publications, 20+ issued and pending patents, gave keynotes and invited lectures at conferences and universities, and frequently received media interviews.