TheSequence Scope: Is Reinforcement Learning Ready for Prime Time

Initially published as 'This Week in AI', a newsletter about the AI news, and the latest developments in ML research and technology

Jun 07, 2020

From the Editor: Is Reinforcement Learning Ready for Prime Time

Reinforcement learning(RL) has become one of the most hyped terms in the machine learning space. Since DeepMind’s AlphaGo defeated Go’s world champion Lee Sedol, RL has been associated with progress in the world of artificial intelligence(AI). After all, a method that can organically learn by trial and error seems to be as close to real intelligence as we can get. The hype around RL is not without merit as RL has been at the center of some of the most important breakthroughs in AI in the last few years. However, most of those applications have been in the field of gaming. Outside that domain, there are serious questions about whether RL is ready for real-world AI scenarios.

The challenges of implementing RL in real-world scenarios take all shapes and forms. From the lack of sophisticated tools and frameworks to the large volumes of data and time required by RL methods, there are plenty of factors that have kept RL constrained to research scenarios or to applications developed by big technology companies. Certainly, democratizing RL development is one of the most important challenges of the next few years of machine learning.

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

AI Research:

Unsupervised Reinforcement Learning

Google Research published two papers proposing the notion of unsupervised reinforcement learning for skill discovery.

The image credit: Google AI

Meta-Graph

Uber Research published a paper proposing Meta-Graph, a new framework for predicting links in graph structures.

Reducing Training Time for Decision Trees

Amazon Research published a paper proposing a method for reducing the training time in decision tree methods without losing accuracy.

Cool AI Tech Releases:

Acme

DeepMind open-sourced Acme, a framework for facilitating reinforcement learning research.

>Read more in the Acme Github page

Enterprise Machine Learning eBook

Algorithmia has published an ebook about the challenges of enterprise machine learning.

AI in the Real World:

Algorithmic Accountability Act

Draft legislation titled the Algorithmic Accountability Act is making the rounds through the United States Congress. The bill urges technology companies to audit AI systems for elements such as bias, discrimination, and other derogatory behaviors.

1touch.io Raises $14 Million

1touch.io is a new startup that helps enterprises to identify repositories that might store or process sensitive data. The company just emerged with stealth mode with $14 million in funding.

Searchable.ai Raises $4 million

Enterprise search startup Searchable.ai for a platform that uses conversational AI to optimize the search experience.

“This Week in AI” is a newsletter curated by industry insiders and the Invector Labs team, every week it brings you the latest developments in AI research and technology.

From July 14th the newsletter will change its name and format to develop ideas of systematic AI education.

To stay up-to-date and know more about TheSequence, please consider to ➡️

TheSequence