Artificial Intelligence #83

Hey, in this issue: Facebook developed a "neural database"; paper explained: Fastformer; Tutorial on inferring concept drift without labeled data; a comparison between the investment returns of business angels and machine learning algorithms; a tutorial on human pose classification; and more.

This issue is sponsored by Bifrost Bridge Studios and Datrics.

• Facebook developed a "neural database" that enables machines to search unstructured data (paper 1, paper 2, code)
• [Paper explained] Fastformer: Additive Attention Can Be All You Need
• [Tutorial] Inferring concept drift without labeled data
• How big data carried graph theory into new dimensions
• [Sponsored] Subscribe to Bifrost Bridge Studios channel and join their Patreon to see exclusive content

The mission of Bifrost Bridge Studios is to educate, uplift, and assist individuals in the rapidly evolving horizon of AI. We focus on familiarizing AI to a diverse audience through various interactive mediums including our animation series on AI, along with our one-of-a-kind graphic novel, Gaia’s Seed.

• A comparison between the investment returns of business angels and machine learning algorithms
• Text data augmentation for deep learning
• [Sponsored] Asking for your feedback for Datrics, probably the best no-code tool to do data science experiments in your browser!
• Hora: an approximate nearest neighbor search algorithm written in Rust with a Python API
• [Tutorial] Human pose classification with MoveNet and TensorFlow Lite

The newsletter is now read by more than 470,000 subscribers. If you are building an AI or a data product or service, you are invited to join the list of these great companies by becoming a sponsor of one of the future newsletter issues. Feel free to reach out to hello@truepositive.ca for more details on sponsorships.

Enjoy the newsletter? Help us make it bigger and better by sharing it with your colleagues and friends.

Have a nice weekend! See you next week. — Andriy