ISSUE 346 · July 27, 2021
In the News
What Ever Happened to IBM’s Watson?
IBM’s artificial intelligence was supposed to transform industries and generate riches for the company. Neither has panned out. Now, IBM has settled on a humbler vision for Watson.
The New York Times | Steve Lohr
Sponsored Link
Understanding Bias in AI: What is it, how to identify it, and how to mitigate it.
It’s no secret...bias in AI is a problem. The more powerful AI capacity becomes, the more important it becomes to ensure that we are producing tools that make life better for more people, not simply scaling and perpetuating forms of error. That’s why we’ve created a guide to understanding and addressing possible sources of bias for your AI/ML project.
Tutorials, Projects & Opinions
BirdNET
BirdNET is a research platform for recognizing bird calls at scale. This is an awesome introduction to the project, including challenges, live demos, and links to free apps. For a look at the machine learning algorithms that drive it, see Machine Learning at Macaulay Library
>>
Cornell Lab of Ornithology | Macaulay Library
Machine-learning on dirty data in Python: a tutorial
There are two Python tutorials here for working with dirty data. The first tutorial shows how to predict missing values and the second shows how to work with non normalized strings.
Dirty data science
Testing Julia: Fast as Fortran, Beautiful as Python
This post walks through a series of tests to compare the performance and code simplicity of Julia, Numpy, and Fortran. There's more to do but the verdict so far is super enthusiastic for Julia.
MatecDev | Martin D. Maas, Ph.D
What is the right level of specialization?
The fragmentation of the data science space seems to be never ending. How much of that specialization is a good thing? When is it bad? This is a short, thoughtful post to help you keep your eye on the ball.
Erik Bernhardsson
30 Days of ML
New to machine learning? Sign up for this new Kaggle challenge to go from beginner to Kaggle competitor in just 30 days. The prerequisites are minimal and it's no cost to join.
Kaggle
Add Vector-Search to Production Applications
Pinecone makes it easy to add vector similarity search to production applications. No more hassles of tuning algorithms or building and maintaining infrastructure. Try it for semantic text search, image/audio search, recommendation systems, and other applications.
// sponsored
Resources
Handbook of Regression Modeling in People Analytics
This new book teaches how to do a wide range of statistical analyses in both R and in Python, ranging from simple hypothesis testing to advanced multivariate modeling. Although it's primarily focused on examples related to the analysis of people and talent, the methods easily transfer to other disciplines. Free to read online.
Keith McNulty
Data Visualization
In defense of simple charts
Simple visualizations don't need to be boring. Here's why, and how, to create great visualizations using simple charting tools. This is a great post that includes lots of examples.
Datawrapper | Lisa Charlotte Rost
100 Days of D3
In this collection of D3 tutorials, Sandra Becker introduces a new chart type every few days with example notebooks, references, and video walk-throughs. Follow along for an easy way to learn D3.
Observable | Sandra Becker
Outlier
Maia, a human-oriented AI for chess
AI-powered chess engines can consistently beat human players. But what if instead of beating humans, an AI engine was trained to understand humans and match its game to a specific player’s playing style? Could it help humans improve their game? Meet Maia, a chess engine that aims to bridge the gap between AI and human chess play.
Microsoft Research
 |