Posted on: April 28, 2019 Posted by: Carly Klein Comments: 0

Today, data is at the forefront of all industry. Companies who collect data can use it to learn about consumer behavior, monetize it by selling it to third parties, interpret it to determine life-altering results, and so much more. The possibilities are endless.

Machine learning (ML) is a method of data analysis and a branch of artificial intelligence technology. The ethos underlying machine learning is the idea that systems can learn from data, identify patterns within that data, and make decisions more efficiently than humans can alone.

Inventors, companies, and developers in the machine learning space should consider patenting their inventions. To do this requires an awareness of the process in obtaining a patent, as well as the difficulties that a complex technology like ML may present in the process.

What is a patent?

A patent is a federal grant of exclusive right given to the owner of an issued patent claim that can last up to twenty years. An issued patent allows the owner to exclude others from making, using, selling or offering an invention covered by the “claims” of the patent.

Patents provide a powerful form of Intellectual Property (IP) protection that can either protect the functionality or the design of an invention. Biotechnology patents fall under the scope of utility patents which cover functionality. Any new and useful process, machine, article of manufacture or composition of matter may qualify for a utility patent.

When an inventor applies for a patent, he or she must demonstrate that the creation meets specific eligibility requirements: an invention must fall under subject-matter eligibility, have utility, novelty, be non-obvious, and not have been previously disclosed to the consuming public.

Machine Learning Patents

Patents seek to protect new and useful methods and inventions that can make existing processes better and more efficient. Machine learning technologies, as methods to analyze, refine, interact with data, are patentable as such. Machine learning is a way of taking the information we have about the interconnectedness of networks and using it to become smarter, creating more efficiency.

Since 1956, innovators and companies have filed nearly 340,000 AI-related patent applications, 40% of which include machine learning technology. Deep learning, a machine learning technique which includes speech and language recognition systems, is the fastest growing AI technology.

Some of the major players in the machine learning patent landscape are Google, Amazon, and Samsung. Since the innate functionality of ML is complex, these companies have invested in research and development to determine how to make ML related technology patent-friendly. One way to do so is by writing and drafting patent claims in such a way that demonstrates a practical usage or an inventive concept beyond what humans can do alone.

What are the complications involved in patenting ML technology?

Machine learning technology may be rejected for patentability under if it fails to meet the requirements of being novel and non-obvious. Technology is often difficult to prove as novel and non-obvious when it can be done by human activity alone, or if it is the exact thing being accomplished by pre-existing technology. If the process accomplished by the technology can be done by a human brain or software alone, it’s not likely to be patentable.

This is where the claim-drafting becomes crucial. In order to be eligible for a patent, ML technology must offer an identifiable improvement, such as making something more effective, efficient, or providing benefit beyond what a human is capable of alone. In the patent claim, if an inventor points to how the technology is a process that creates more efficiency or provides a new method to analyze data that hasn’t been seen before, it can be patented.

Tips for ML inventors

For an inventor with an ML product, when applying for a patent, he or she inventor must show in the list of claims that the software is a process that has novelty, non-obviousness, and utility. The claims must make it apparent that the technology can’t be done by humans alone and that it brings usefulness to human life that was nonexistent prior to the technology’s inception.

Focus on the functionalities of the ML technology, device or system that you want to protect and let your patent attorney handle the details of patent law that dictates how this is done.

In the list of claims, be sure to reference the hardware components and physical devices involved in making the ML technology operate. Additionally, demonstrate the benefit that the invention will bring to humanity. Show how the invention is saving time, saving space, increasing multi-party communications, increasing efficiency, etc.

It is also wise to do a patent search before starting the patent process to ensure someone else hasn’t already tried patenting a similar technology. Do so quickly, because technological innovation moves fast. It is best to capitalize on your invention before the market changes or another company develops something so similar to your invention that you’re barred from a patent.

Carly Klein is a first-year student at Loyola Law School. A Los Angeles native and a graduate from Boston University with a B.A. in Political Science & Philosophy, she seeks to pursue a career in civil litigation.

Leave a Comment