Last Week in AI

Source: Deep Learning on Medium

Every week, Invector Labs publishes a newsletter that covers the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:

From the Editor: Streamlining AI Adoption in 2020

This would be the last newsletter of this year as we will be taking a small break for the holidays and, quite frankly, so are most artificial intelligence(AI) researchers. When I think about my AI hopes for 2020, I don’t think as much about exoteric research milestones as much as I do about practical technical achievements. I believe in the next few years there should be a strong focus about simplifying the implementation of AI solutions.

Despite the undeniable achievements of AI in the last few years, the implementation of AI solutions remains a paramount challenge for most organizations. Complex infrastructure, data curation, a difficult lifecycle and a very limited access to talent are some of the factors that contribute to make AI a high barrier of entry for most organizations. To not widen the gap between AI research and its practical applications, the industry needs to enable mechanisms that streamline the adoption of AI solutions. We are certainly working on it at Invector Labs.

I hope you have enjoyed this newsletter during 2019 and I would like to wish you a very blessed Holiday season.

Now let’s take a look at the core developments in AI research and technology this week:

AI Research

Neural Teaching Networks

Uber researchers published a paper introducing the concept of Generative Teaching Networks(GTNs) which are able to generate data or environments to train other networks.

>Read more in this blog post from the Uber engineering blog

Improving Distributed Reinforcement Learning

Microsoft Researchers published a paper proposing a method that helps distributed reinforcement learning agents to select learning targets automatically.

>Read more in this blog post from Microsoft Research

AI for the Speech Impaired

DeepMind published a blog post detailing how they use WaveNet models to generate natural voices for speech impaired users.

>Read more in this blog post from DeepMind

Cool AI Tech Releases

Open Sourcing ALBERT

Google open source ALBERT, a self-supervised learning model for state of the start language analysis tasks.

>Read more in this blog post from Google Research

Resemble AI

Startup Resemble AI launched a voice synthesis platform and deep fake detection tool.

>Read more in this coverage from VentureBeat

AI in the Real World

SoftBank Chief Wants to Make AI Exams Mandatory

Masayoshi Son said that Japan should make artificial intelligence (AI) a mandatory subject.

>Read more in this coverage from Nasdaq.com

Recognizing the Absence of Things

A major milestone for deep learning might be to recognize the absence of objects.

>Read more in this article of IEEE Spectrum

AI’s Role in Warfighting

Lt. Gen. Jack Shanahan is the new AI Chief of the Pentagon and he has a pragmatic view of AI in warfighting.

>Read more in this coverage from Wired