[ML UTD 1] Machine Learning Up-To-Date

Join LifeWithData in Machine Learning Up-To-Date 1 for a curated, concise machine learning newsletter that points the reader to key developments.

Welcome to Machine Learning Up-To-Date (ML UTD) 1! The LifeWithData blog separates the signal from the noise in today’s hectic front lines of software engineering and machine learning.

LifeWithData aims to consistently deliver curated machine learning newsletters that point the reader to key developments without massive amounts of backstory for each. This enables frequent, concise updates across the industry without overloading readers with information.

Machine Learning Up-To-Date 1 brings innovations in the fields of edge computing, deep learning, ML standardization.

Edge Computing

Let’s continue moving away from bulky cloud server costs with Pytorch mobile and SwiftUI. Cloud computing costs rise very quickly when computationally-expensive ML models come into play. Moving this work to the edge device will bring many benefits.

WebAssembly enables Google’s MediaPipe to climb into browsers from Desktops. This library didn’t receive as much popularity as I imagined, but it could certainly change soon.

Deep Learning

Thinc-ing the end could be in sight for enterprise machine learning framework wars. Check these guys out if you are tired of “Pytorch vs Tensorflow” message threads. They put the icing on the cake with great documentation and examples.

“Computer, please merge into the middle lane for me” — Apple. The tech giant has demonstrated teaching an AI agent to effectively merge lanes while driving (in a simulator).

“Hey Alexa, tell Siri that Google is making a new AI agent, Meena, to deliver a truly conversational agent to customers” Making a conversational AI agent remains elusive even to big data and deep learning, but Google wants to change that.

Microsoft’s Project Tokyo aims to give sight to the blind! This new project combines augmented reality with machine learning to help visually impaired users gain a sense of sight through audio-based feedback.

ML Standardization

Literally the Google for Datasets is out of beta. I am in love.

Stanford’s DAWNBench formally makes way for MLPerf. The ability for one software framework to publicly step aside for a preferred one speaks to the true nature of open-source software.

Wrap Up

What do you think about this format? Do you like seeing more links, with less detail, or vice versa?

What would you like to see done differently with this newsletter?

What do you think is missing from other newsletters you follow?

Originally published at https://lifewithdata.org on January 31, 2020.

Living the LifeWithData.org | Top writer in AI