At the Deep Learning Summit in San Francisco earlier this year, RE•WORK brought together global experts in AI and Deep Learning to explore the latest breakthrough and cutting edge topics in the space. Both industry and academia were represented and topics ranged from education and encouraging the future generations into AI, to technical progressions in neural networks. At the summit there were 10 stages honing in on specific topics across the two-day event. Video presentations and workshops are now available for the following stages: Deep Learning, AI Assistants, Futurescaping, Ethics & Social Responsibility, and Industry Applications. Take a look at some of the presentations and panel discussions:
Go-Explore: A New Type of Algorithm from Hard-exploration Problems
Jeff Clune, Senior Research Scientist & Founding Member, Uber AI Labs
A grand challenge in reinforcement learning is producing intelligent exploration, especially when rewards are sparse or deceptive. I will present Go-Explore, a new algorithm for such ‘hard exploration problems.’ Go-Explore dramatically improves the state of the art on benchmark hard-exploration problems, enabling previously unsolvable problems to be solved. I will explain the algorithm and the new research directions it opens up. I will also explain why we believe it will enable progress on previously unsolvable hard-exploration problems in a variety of domains, especially the many that harness a simulator during training (e.g. robotics). More information can be found at https://eng.uber.com/go-explore.
Conversational AI @ Facebook
Rushin Shah, Engineering Leader at Facebook
Conversational AI is becoming a key part of various products at Facebook, including Portal, Messenger suggestions, Marketplace, recommendations, to name just a few. Being able to ship delightful conversational experiences means we also need to invest in solving hard research problems in areas such as language understanding, dialog management and language generation. In this talk, I will talk about some of the current work we're doing in these areas, as all as work we've done with Pytorch in order to enable our research scientists to quickly prototype and deploy advanced NLP models.
Breaking Through Challenges in Industry Series: OCR at Scale at Dropbox
Thomas Berg, Machine Learning Engineer at Dropbox
There are many billions of images in Dropbox. About 15% of them are photos of documents -- receipts, business cards, contracts, etc. -- with text content that is hidden from our search index. To allow our users to search for these "documents," Dropbox built an OCR system. In this talk I'll describe Dropbox's OCR project from its initial, small-scale deployment of a third-party library, through development and large-scale deployment of a homegrown solution using deep networks, focussing on the performance and scaling problems we encountered and solved along the way.
Keen to watch more videos from the summit? Check out the library of videos here.