When you see the list of patents that DeepMind has filed you will be shocked and not in awe. It reinforces not learning, but the idea that all software patents are stupid.

I/We are indebted to the wonderfully named IPKat blog which deals in IP (intellectual property not Internet Protocol) issues for a list of some of the patents that DeepMind is filing. Just read the titles and be appalled.

• WO 2018/048934, "Generating Audio using neural networks", Priority date: 6 Sep 2016
• WO 2018/048945, "Processing sequences using convolutional neural networks", Priority date: 6 Sep 2016
• WO 2018064591, "Generating video frames using neural networks", Priority date: 6 Sep 2016
• WO 2018071392, "Neural networks for selecting actions to be performed by a robotic agent", Priority date: 10 Oct 2016
• WO 2018/081089, "Processing text sequences using neural networks", Priority date: 26 Oct 2016
• WO 2018/083532, "Training action selection using neural networks", Priority date: 3 Nov 2016
• WO 2018/083667, "Reinforcement learning systems", Priority date: 4 Nov 2016
• WO 2018/083668, "Scene understanding and generation using neural networks", Priority date: 4 Nov 2016
• WO 2018/083669, "Recurrent neural networks", Priority date: 4 Nov 2016
• WO 2018083670, "Sequence transduction neural networks", Priority date: 4 Nov 2016
• WO 2018083671, "Reinforcement learning with auxiliary tasks", Priority date: 4 Nov 2016
• WO 2018/083672, "Environment navigation using reinforcement learning", Priority date: 4 Nov 2016

This list covers a very short time span September to November 2016 and any since December 2016 are still in the pipeline.

The titles are enough to indicate that these patents should not have been filed and certainly should not be granted, but take a look at the abstract from the one on reinforcement learning.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for prediction of an outcome related to an environment. In one aspect, a system comprises a state representation neural network that is configured to: receive an observation characterizing a state of an environment being interacted with by an agent and process the observation to generate an internal state representation of the environment state; a prediction neural network that is configured to receive a current internal state representation of a current environment state and process the current internal state representation to generate a predicted subsequent state representation of a subsequent state of the environment and a predicted reward for the subsequent state; and a value prediction neural network that is configured to receive a current internal state representation of a current environment state and process the current internal state representation to generate a value prediction.

Yes it may be legalese, but you should be able to see that this covers all reinforcement learning - probably even B F Skinner's box, let alone Pavlov's dogs. Notice it doesn't say Artificial Neural Network so wetware might well be included.

The patents haven't been granted yet, and it is clear that they shouldn't be. In the past Google has applied for patents on AI and has said that they are "defensive", i.e. an attempt to stop others from registering and misusing them. This is noble, but not really the answer - why trust Google or any company to defend the subject.

The real solution is for this sort of things to not be patentable - there is so much prior art that it really should not pass.

More Information

DeepMind: First major AI patent filings revealed

Related Articles

Google Files AI Patents

Big Increase in AI, Cognitive and Cloud Computing Patents in 2016

US Appeals Court Divided Over Software Patent Eligibility

The Oracle v Google Trial IProgrammer Reads the Patents

Patent Wars

Patent Evil

Find Prior Art Added to Google Patent Search

White House Advises That APIs ARE Copyrightable       

Fujifilm Has A Patent On Converting To Greyscale       

Google Frees Up More Patents  

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info