• George Data Science

Google Launches Reinforcement Learning Frameworks to Train AI Models

Google, the American multinational technological company, has brought out an open source framework in connection with reinforcement learning (RL) based on Tensor Flow.

Tensor Flow is the machine learning library of Google. According to the members of the Google Brain team, the development will enhance stability, reproducibility and flexibility – three of the prominent attributes of mind.

Reinforcement learning is a branch of machine learning which involves actions to maximize the outcomes of rewards related to a certain element. The technical learning has been through a transition over the years to reach its current stage. Deep mind – a startup in London which was acquired by Google later on – is one of the classical working examples of this kind of AI.

DQN, an RL project of Deep mind, has the feature to play games in a way which surpasses the ability of even the expert players among human beings.

Challenges involved in building reinforcement learning systems

The iteration of designs is one of the most essential elements for the development of RL systems. Sometimes it also involves the disruption of established frameworks without an obvious path to work out a solution. The only catch is that some of the pitfalls in its methodology can slow down the progress of the process and limit exploration at the subsequent stages.

Most experts believe the development is a major breakthrough in the overall RL frameworks as it brings about the combination of flexibility and stability – a trait which is not available in any of the existing variants of frameworks that are in use at present.

Furthermore, producing the exact copies of frameworks from the existing units can be a tedious process as it takes away a lot of time. If adopted, such a practice can result in scientific reproducibility issues in future.

Benefits linked to the release of the open source framework

The package from Google comes across with some obvious benefits. Among other benefits, it is expected to highlight the importance of reproducibility by virtue of the use of certain tools. This, according to experts, will help researchers benchmark their ideas against established methods at a brisk pace.

For those who are testing the waters in terms of realizing the fresh avenues in RL, the move would make their life easy by saving their valuable time. And time, as we all know, is more valuable than money.

In the words of Marc Bellemare and Pablo Samuel Castro, the prominent Google researchers “Inspired by one of the main components in reward-motivated behavior in the brain, and reflecting the strong historical connection between neuroscience and reinforcement learning research, this platform aims to enable the kind of speculative research that can drive radical discoveries”.

They are hopeful about the role of the flexibility and the ease of use of the framework will go a long way in encouraging and empowering researchers to try out new radical and incremental ideas.

While the new development is expected to help people iterate over several ideas in quick time, it remains to be seen how the members of the larger community utilize it to their advantage in the next few months down the line.


A Unit Of The George Telegraph Group

Quick Links

Our Programs

Contact Us For Data Science Training

Office No.: 804D
Godrej Waterside Building, Tower 1, 8th Floor, 
Sector V, Block DP,
Kolkata - 700091

  • Facebook
  • LinkedIn - Black Circle

2019 George Institute of Data Science All rights reserved