#10 Paris Women in Machine Learning & Data Science: Applied ML to Finance, Physics & Tech Journalism
Our 10th meetup took place on the 29th of November 2018 at leboncoin! We were extremely pleased to welcome many newcomers for this session.
We launched the meetup with a few announcements:
1.🌎The WiMLDS community is growing. New chapters in Madrid, London, Montréal, Tokyo recently opened and Paris counts now more than 1917 members! The Paris chapter is the third biggest one of the WiMLDS community.
💻Actu IA is looking for contributors to write in French about Artificial Intelligence. Feel free to reach out to them if you are interested! Our very own Caroline Chavier recently introduced the Paris WiMLDS meetup to Actu IA during France is AI:
📖4 attendees of the Paris WiMLDS meetup won Chloé-Agathe Azencott’s book: Introduction au Machine Learning! Thanks to all of you who took part to the online and IRL contests!
💌 Last but not least, we launched our very first survey to better understand your needs and organize in consequence our future meetups. If you have 4 minutes to spare us, we would be grateful: bit.ly/2Fbehbb 📌
📅 Here is a look at the 2019 calendar of our future meetups! As usual, if you want to speak (or know somebody that wants to speak), drop us a line at firstname.lastname@example.org !
Our opening talk for the evening was from Elise Tellier & Marine Michaut from Capital Fund Management, with “Applied Machine Learning in Finance: Estimating Missing Bid-Ask Spreads and Detecting Anomalies”.
The presentation focused on Data Science in the finance world, where data may be hard to come by, and have significant missing entries. They took us through their journey of starting with data analysis, continuing with filling missing entries in time series of financial spreads, and ending with model crafting and selection. Their advice is: start simple, and look carefully at your data! Sometimes the simplest models performs much better than the latest hype for your dataset.
Our second talk was from Lenka Zdeborova, a former physicist turned ML researcher, on “Statistical Physics Studies of Machine Learning Problems”.
Her talk dived into the fascinating details of connecting two apparently separate fields as physics and machine learning with the use of phase transitions. As with physics, one can start with a very simple data generation model to assess theoretical guarantees. The fundamental question she assessed was : for a given algorithm and a given number of samples what is the optimal error that can be achieved? Using the link with statistical physics she showed regions of parameters where an algorithm achieves the optimal performance and locates sharp phase transitions separating learnable and non-learnable regions. One of her study cases involved a model with few hidden units.
The conclude the evening, our third talk by Lucie Ronfaut focused on how to become a tech journalist; focusing on what cliches and what problems one must address in order to proceed in this career. From how to define tech, how to avoid clickbaits, the different types of stories to write with different approaches, Lucie explained with irony how difficult yet satisfying this career is.
🎊 🎉 As usual, we concluded with a networking session, with great food and drinks.
If you want to keep track about what’s next, you can :
📩send an email to the Paris WiMLDS team to keep in touch >email@example.com
📍join our Slack channel for more discussions about machine learning, data science, and diversity in tech
📑check our Google spreadsheet if you want to speak 📣, host 💙or help 🌠!