Using AI to control AI: How to Prevent Creating Biased Datasets

Author(s): Michelangiolo Mazzeschi

Originally published on Towards AI.

What are labels?

In the last few days, MIT took down a cited 80 million tiny images 32×32 size because it contained labels (if you do not know what it means, I will clarify it further on) that were inappropriate. In this article, I will attempt to tackle this specific problem using a Machine Learning approach.

If I understood correctly from the published paper on this dataset, this was the procedure undertaken in collecting data from the internet and storing it into the dataset:

Retrieving images in large scale from Google

The team responsible for the project scavenged the internet for these images, each one associated… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI