Issue #259

October 9, 2024

Welcome to the October 9th edition of the Data Science Briefing!

The next edition of the Generative Artificial Intelligence with the OpenAI API for Developers is coming up in just a few hours, but you can still Register and not miss out!

Following the great success of the first edition of the LLMs for Data Science webinar that took place on Sept 19th, we’re proud to announce the second edition coming up on Dec 13th. Last time, we ran out of spots, so Register now so you don’t get left out!

The most recent post on the Epidemiology series, Epidemiology 303: Metapopulation Models, explores how to connect multiple populations through a travel matrix. In the graph subsection, you’ll find all you need to know about k-core Decomposition, while in the Visualization section, you can explore The Effects of Vaccination through a WSJ visualization.

In the five years I’ve been writing this newsletter, I’ve never had a chance to cover the various Nobel prizes as they were always off topic. This all changed this year, with not one, but two Nobel prizes awarded to developments in AI. The 2024 Nobel Prize in Physics was awarded to Hopfield Networks and Boltzmann machines, two off-shoots of Sherrington-Kirkpatric spin-glasses that led to the eventual delopment of deep-learning networks. The 2024 Nobel Prize in Chemistry was awarded to AlphaFold, the deep-learning system that (essentially) solved protein folding, a problem that was thought unsolvable not that long ago.

In our non-Nobel-related content, we have an overview of how Hopfield Networks is All You Need, an essay On the Nature of Time by Stephen Wolfram, thoughts on LLMs, Theory of Mind, and Cheryl’s Birthday and a look at how diffusion models work: the math from scratch.

On the academic front, we explore the Unsupervised detection of coordinated fake-follower campaigns on social media, Differential Transformers, Contextual Document Embeddings, and the Emergence of social phases in human movement.

This week’s book recommendation is “Why Machines Learn: The Elegant Math Behind Modern AI” by A. Ananthaswamy. You can find all the previous book recommendations on our website. In this week’s video, we have a tutorial on Object Counting with CNNs.

Data shows that the best way for a newsletter to grow is by word of mouth, so if you think one of your friends or colleagues would enjoy this newsletter, go ahead and forward this email to them. This will help us spread the word!

Semper discentes,

The D4S Team

This week's book is "Why Machines Learn: The Elegant Math Behind Modern AI" by A. Ananthaswamy. The book introduces the main ideas and developments of Artificial Intelligence clearly and concisely. Starting with the invention of the Perceptron in the 50s, through all the significant developments of the last several decades, such as Support Vector Machines, Hopfield Networks, and Backpropagation, to the latest developments in Large Language Models. Ananthaswamy explains how they fit in the historical development of Computer Science and AI, as well as how they connect to insights originating in biology and psychology.

The book targets a general audience familiar with basic math. Mathematical concepts such as probability and linear algebra are introduced in an intuitive way that provides just enough detail to understand the more technical parts of the text. Overall, a great resource whether your reviewing these concepts or encountering them for the first time.

Why Machines Learn: The Elegant Math Behind Modern AI

1. 2024 Nobel Prize in Physics [nobelprize.org]
2. 2024 Nobel Prize in Chemistry [nobelprize.org]
3. Hopfield Networks is All You Need [ml-jku.github.io]
4. On the Nature of Time [writings.stephenwolfram.com]
5. An Intuitive Explanation of Black–Scholes [gregorygundersen.com]
6. LLMs, Theory of Mind, and Cheryl's Birthday [github.com/norvig/]
7. Introducing canvas [openai.com]
8. How diffusion models work: the math from scratch [theaisummer.com]

• Overcoming bias in estimating epidemiological parameters with realistic history-dependent disease spread dynamics (H. Hong, E. Eom, H. Lee, S. Choi, B. Choi, J. K. Kim)
• Unsupervised detection of coordinated fake-follower campaigns on social media (Y. Zouzou, O. Varol)
• Reconstructing networks from simple and complex contagions (N. W. Landry, W. Thompson, L. Hébert-Dufresne, J.-G. Young)
• Emergence of social phases in human movement (Y. Zhang, D. Sarker, S. Mitsven, L. Perry, D. Messinger, U. Rudolph, M. Siller, C. Song)
• Differential Transformer (T. Ye, L. Dong, Y. Xia, Y. Sun, Y. Zhu, G. Huang, F. Wei)
• Contextual Document Embeddings (J. X. Morris, A. M. Rush)

Object Counting with CNNs (Intro to Computer Vision Part 1)

October 9, 2024

​