Episode 1: Intrusive Face Detection, Kaggle Cheaters, AlphaFold, and Becoming an A.I. Researcher

A4N (AI/Machine Learning News)

For our inaugural episode of A4N — the "Artificial Neural Network News Network", a lighthearted podcast covering A.I. advances — we discuss real-time face recognition by London police, cheating in the famed Kaggle data-science competition, the landmark AlphaFold model for predicting protein structure from DNA, and how to become (or at least hire!) an A.I. researcher.

Below’s a detailed breakdown of the episode’s four segments, with time stamps for all of the references that we mentioned.

Segment 1 — Global headline news: Intrusive real-time face recognition

In this segment, we discuss a controversial new approach by the Met police force in London, which is to use a (low-performing) facial-recognition system to flag “known criminals” in real-time in train stations.

Segment host: Vince Petaccio II

Reference news article from The Guardian (1:10)

Citizen App (9:10)

Segment 2 — “Sports”: Kaggle Cheating

In this segment, we introduce what the Kaggle data-science competition platform is and how folks (now formerly!) working at the well-known firm H2O.ai cheated to perform well. How unsportspersonlike!

Segment host: Andrew Vlahutin

Reference news article from Towards Data Science (14:38)

Segment 3 — Health: AlphaFold

In this segment, we introduce how DNA encodes proteins that do all of the work in our bodies. We then describe the new AlphaFold algorithm that crushes all of the existing approaches at predicting protein structure from DNA-sequence data. In the benchmark “CASP” competition, AlphaFold correctly predicted the structure of 58.1% of the proteins while the second-best algorithm correctly predicted 7.0% of them.

Segment host: Grant Beyleveld

Reference blog post from DeepMind (26:31)

CASP (30:05)

AlphaGo documentary film (37:20)

Segment 4 — Classifieds: How to become (or hire!) an A.I. Researcher

In this segment, we list ways that you can find hidden-gem A.I. researchers to hire from within a high-demand field. We also list approaches for breaking into the field of A.I. if you come from a non-traditional background, e.g., you don’t already have a PhD in machine learning or statistics.

Segment host: Jon

How to Hire Smarter than the Market (38:33)

Getting Hired in AI as Self-Taught Researcher (40:39)

Deep Learning book by Goodfellow et al. (41:30)