We had an incredible time listening to all of the fantastic speakers at the RE•WORK San Francisco Summit in January. We are incredibly thankful for the diverse range of topics covered across the AI/DL spectrum from leading industry experts. Whilst every single AI presentation delivered incredible insight, AI concepts & research, we have narrowed it down to five presentations we think you have to see (with a few extras at the end...shh).
Machine Learning - A New Approach to Drug Discovery - Daphne Koller, Founder, Insitro
As the CEO and Founder of Insitro, Daphne and her team have been working on deploying Machine Learning at various stages of drug development to predict any experimental outcomes that would otherwise cost both time and resources. This research is incredibly important as it becomes consistently more challenging to develop new medicine. In this insightful presentation, Daphne talks us through how she leverages the phenotyping/CRISPR and Machine Learning revolutions to create the best possible tool to elevate the 5% success rate in pharmaceutical research & development. Watch Daphne's presentation here.
Understanding the Behaviour of Series Data Using The Matrix Profile And Deep Learning - Frankie Cancino, Data Scientist, Target
Frankie Cancino took the stage and gave us insight into what goes on under the surface at Target. He shared some of the projects he is currently working on including the use of data streams to forecast consumer behaviour, which also comes into play as an operational tool to help with detecting data anomalies. Frankie covered a fascinating range in his presentation including python libraries, matrix profiles & data transformation, whilst also uncovering Target’s plans for the future including GPU Implementation, unifying data libraries and API redesign. Watch Frankie's presentation here.
Enhance Recommendations in Uber Eats with Graph Convolutional Networks - Ankit Jain & Piero Molino, UberAI
Ankit Jain and Co-Founder Piero Molino are both Senior Data/Research Scientists at Uber AI. With a specific focus on Uber Eats, they gave an excellent presentation on how to augment the ranking models to show better representations of dishes and restaurants for users. They took us through 3 important steps in their process: Graph Representation Learning, Dish Recommendation on Uber Eats & Graph Learning. You can watch their full presentation here.
Multi-modal Video Content Analysis for Content Recommendation - Shalini Ghosh, Principal Scientist & ML Research Leader, Samsung Research America
Shalini delivered a great presentation talking us through the steps of Multi-model video content analysis. From generating fine-grained categories through to the training of baseline models using inception and DNN, Shalini didn’t leave much to the imagination! Shalini makes an interesting point when distinguishing between the visual cues of humans and machines. Watch the full presentation for detailed explanations of Temporal Coherence in NetVLAD and RNN models, experiments with Hierarchal Loss and Fusion Architecture and any future work at Samsung! You can see Shalini's talk here.
Fireside Chat - Lex Fridman & Ilya Sutskever
It was so exciting to have Lex Fridman and Ilya Sutskever share the stage. The two AI experts went into detail on the fundamental mechanisms behind OpenAI’s current projects & successes: GPT-2 Language Modelling, the Reinforcement Learning that drove DOTA’s defeat by OpenAI Five and music generator MuseNet. Have a look at the full video for more information on Transformers, Self Play, Automated Curriculum and what the future holds in this line of work. A brilliant and insightful discussion. Watch it here.
Additional Content....
I know we said five videos... but there was quite a lot of content we think you need to see. So here are three more videos from the first few months of 2020!
Saving the Rainforest with AI - Topher White, CEO of Rainforest Connections
20 Questions with OpenAI's Research Team Lead, Jeff Clune
The Future of FAIR (Facebook AI Research) with Michela Paganini
Interested to get first access to more content? Subscribe to our newsletter for our bi-weekly newsletter! Register here.