Nov 18, 2020

6 Posts

Detecting system pointing out scratches on a surface
Nov 18, 2020

The Batch: Bias In Surprising Places, Retail Models Adjust to Covid, Faster Transformers, AI Patents Explode

Last week, I wrote about the limitation of using human-level performance (HLP) as a metric to beat in machine learning applications for manufacturing and other fields. In this letter, I would like to show why beating HLP isn’t always the best way to improve performance.
Collage of self portraits
Nov 18, 2020

Unsupervised Prejudice

Social biases are well documented in decisions made by supervised models trained on ImageNet’s labels. But they also crept into the output of unsupervised models pretrained on the same dataset.
Figures related to retailers' sales during the pandemic
Nov 18, 2020

Retailers Adjust to the Pandemic

Covid-19 wreaked havoc with models that predict retail sales — but China’s biggest annual e-commerce event showed that they’re back in business.
Data related to Nvidia's Pay Attention When Required (Par) approach
Nov 18, 2020

Selective Attention

Large transformer networks work wonders with natural language, but they require enormous amounts of computation. New research slashes processor cycles without compromising performance.
Data related to AI patents in the U.S.
Nov 18, 2020

Charting the AI Patent Explosion

A new study used AI to track the explosive growth of AI innovation. Researchers from the U.S. Patent and Trademark Office deployed natural language processing to track AI’s increasing presence in four decades of patent data.
Detecting system pointing out scratches on a surface
Nov 18, 2020

AI Versus Human-Level Performance, Part 2

Last week, I wrote about the limitation of using human-level performance (HLP) as a metric to beat in machine learning applications for manufacturing and other fields. In this letter, I would like to show why beating HLP isn’t always the best way to improve performance.

Subscribe to The Batch

Stay updated with weekly AI News and Insights delivered to your inbox