Cloud Hosting

To print this article, all you need is to be registered or login on Mondaq.com.

Note: First published inThe Intellectual PropertyStrategistandLaw.com.

This article is Part One of a Three-Part Article Series

Artificial intelligence is changing industry and society, andmetrics at the US Patent and Trademark Office (USPTO) reflect itsimpact. In a recent publication, the USPTO indicated that from 2002to 2018 the share of all patent applications relating to artificialintelligence grew from 9% to approximately16%.SeeInventing AI, Tracing thediffusion of artificial intelligence with U.S. patents,Office of the Chief Economist, IP Data Highlights (October 2020).For the foreseeable future, patent applications involvingartificial intelligence technologies, including machine learning,will increase with the continued proliferation of suchtechnologies. However, subject matter eligibility can be asignificant challenge in securing patents on artificialintelligence and machine learning.

This three-part article series explores USPTO handlingofAliceissues involving artificialintelligence and machine learning through a sampling of recentPatent Trial and Appeal Board (PTAB) decisions.See AliceCorp. v. CLS Bank Int'l, 134 S. Ct. 2347 (2014). Somedecisions dutifully applied USPTO guidelines on subject mattereligibility, including Example 39 thereof, to resolve appeal issuesbrought to the PTAB. In one case, the PTABsuasponteoffered eligibility guidance even withnoAliceappeal issue before it. These decisionsinform strategies to optimize patent drafting and prosecution forartificial intelligence and machine learning relatedinventions.

Generic Machine LearningAlgorithm

InEx parte Hussain, Appeal No. 2020-005406 (PTABFeb. 18, 2021), the PTAB considered the subject matter eligibilityof claims reciting a machine learning algorithm inrelation to mitigation of risk of consumer default on an onlinetransaction. Representative claim 1 recited as follows:

Id.at 2-3 (emphasis added). To assess subjectmatter eligibility of the representative claim, the PTAB appliedUSPTO guidelines mandating the familiar two step analyticalframework.SeeUSPTO, 2019 Revised PatentSubject Matter Eligibility Guidance, 84 Fed. Reg. 50 (Jan. 7,2019); USPTO, October 2019 Update: Subject Matter Eligibility, 84Fed. Reg. 55942 (Oct. 17, 2019).

As to the first prong of Step 2A in the analytical framework,the PTAB indicated that the representative claim used only ageneric machine learning algorithm to output afidelity score in some unspecified manner. The PTABalso addressed Example 39 of the USPTO guidelines, the examplereciting machine learning in a hypothetical claim deemed eligible.In particular, the PTAB contrasted relevant detail in the claim ofExample 39 versus the relative absence of such detail in therepresentative claim. The PTAB acknowledged that the representativeclaim expressly recited that the machine learning algorithm wastrained to infer characteristics about a user from variable valuesgenerated from specific types of data. Nonetheless, the PTABreiterated that the machine learning algorithm as claimed wastrained to make inferences in an unspecified way without anytechnical details. The PTAB gave little consideration to therecited transformation of the specific types of data into thevariable values, which were specifically claimed as inputs to themachine learning algorithm. Depending on the facts, the claimedinputs to the machine learning algorithm could have been deemedsuggestive of data to train the machine learning algorithm. Forthat reason, the claimed inputs might have been argued topotentially resemble or parallel the recitation of training datadetails supporting eligibility in Example 39. However, no sucharguments were raised.

The PTAB found that a machine learning algorithm assuch was not described in the specification despitethe acknowledged references in the specification to a logisticregression, random forest, supervised learning algorithm, neuralnetwork, vector machine, and other classification algorithm.According to the PTAB, the description of these otherconcepts without technical details confirmed the abstractnature of the claimed machine learning algorithm. In particular,the PTAB noted that the specification described algorithms togenerate fidelity scores without details of trainingthem to infer characteristics about users. Refusing to alsoconsider the machine learning claim limitations under the secondprong because they recited the abstract idea under the first prong,the PTAB ultimately determined that the representative claim wasineligible after finding no inventive concept.

Accordingly, not just any claimed specifics about an artificialintelligence related invention will satisfy the PTAB abouteligibility. Although the representative claiminHussainrecited the inference specificallygenerated by the machine learning algorithm, the PTAB indicatedthat the claim still did not specify enough. In view of thePTAB's observation that both the specification andrepresentative claim lacked technical detail, expressly claimingtraining data and identifying it as such and of coursebeforehand drafting the patent application in support thereof might have secured a different outcome.

Part Two of this article series will further analyze recent PTABdecision making regarding artificial intelligence and subjectmatter eligibility.

The content of this article is intended to provide a generalguide to the subject matter. Specialist advice should be soughtabout your specific circumstances.

POPULAR ARTICLES ON: Intellectual Property from United States

Frankfurt Kurnit Klein & Selz

Another bombshell has dropped. Is change coming? For over a decade, it was generally considered safe to include an image or a video on your website that was linked through to the social media...

Obhan & Associates

Trademarks Comparative Guide for the jurisdiction of India, check out our comparative guides section to compare across multiple countries

Finnegan, Henderson, Farabow, Garrett & Dunner, LLP

In Omni MedSci, Inc., v. Apple Inc., Nos. 2020-1715, -1716 (Fed. Cir. Aug. 2, 2021), a divided Federal Circuit panel affirmed the district court's denial of Apple's motion to dismiss.

Husch Blackwell LLP

Aside from the regulatory requirements imposed on beer labels, as discussed in the Anatomy of a Beer Label: Part I post on COLAs, brewers should consider protecting the trademarks featured on their beer labels.

See the article here:
Artificial Intelligence And Subject Matter Eligibility In US Patent Office Appeals Part One Of Three - Intellectual Property - United States - Mondaq...