Abu Dhabi Machine Learning Season 1 Episode 5
08.06.2021 - Abu Dhabi Machine Learning - ~2 Minutes
When?
- Tuesday, June 8, 2021 from 7:00 PM to 9:00 PM (Abu Dhabi Time)
Where?
- At your home, on zoom. All meetups will be online as long as this COVID-19 crisis is not over.
Programme:
Talk 1: Deep Learning for breast cancer screening; From Research to Production
Pierre Fillard and team.
Therapixel https://www.therapixel.com/
Organizer’s note (Gautier Marti): My Machine Learning journey really started with those guys (and with Olivier Clatz in particular); At the time they were still part of INRIA (team: Asclepios), a French top-tier academic research institution in applied mathematics and computer science. They spun-off a couple of months later into Therapixel.
Talk 2: Towards more informative embeddings with metric learning
Abstract:
Embeddings have become a crucial tool in machine learning, enabling the transformation of inputs from a high-dimensional, difficult-to-interpret space to an easier-to-use representation. In quant research, they enable the representation of complex unstructured data (e.g., 10K reports) in ways amenable to predictive models. However, embedding design is an exercise in trade-offs: only some aspects of the input space are captured in an embedding and it’s often not clear what those are. Metric learning, the subfield of machine learning dedicated to studying distance measures, provides an elegant and powerful tool to investigate embeddings. With examples from computational genomics and natural language processing, we describe how metric learning can be applied to design more informative embeddings by identifying, aggregating, and accentuating their most informative features.
Speaker Bio:
Rohit Singh is co-founder and CTO at martini.ai - a startup generating deep learning alpha for fixed income. Rohit is also a Research Scientist at MIT and is affiliated with the Computer Science and Artificial Intelligence Lab there. Rohit was the CEO of Tech Square Trading, a quantitative hedge fund for six years. Using signals generated by statistical and machine learning methods, the firm traded about $285MM across the US, Europe and Japan each day. Previously, he was at Cubist Systematic, the quant equity sub-division of SAC Capital/Point72. He is the recipient of MIT’s George M. Sprowls Award for Best CS PhD Theses and of Stanford’s Christopher Stephenson Memorial Award for Masters Research in CS. His LinkedIn profile is https://www.linkedin.com/in/rohit-singh-0b509b2/.