Recentive Analytics, Inc., v. Fox Corp., Appeal No. 2023-2437 (Fed. Cir. Apr. 18, 2025)
In our Case of the Week, the Federal Circuit addressed a question of first impression concerning whether developments in machine learning are subject to patenting. The Court found that “the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.”
At issue, the machine learning training patents at issue generally recite the functions of collecting the data, iteratively training the model, generating an output, and retraining the model based on new data inputs. The specification also recites “any suitable machine learning technique[,] . . . such as, for example: a gradient boosted random forest, a regression, a neural network, a decision tree, a support vector machine, a Bayesian network, [or] other type of technique” may be used. The network mapping patents uses machine learning to generate optimized network models.
The district court dismissed the case because it had concluded that, under the Alice test, a two-step legal framework, the patents-in-suit were directed to an abstract idea and there was not significantly more to give the abstract idea an inventive concept. It found that network maps were abstract ideas and were generated using a generic mathematical formula. Furthermore, the court found that “the machine learning limitations were no more than ‘broad, functionally described, well-known techniques’ and claimed ‘only generic and conventional computing devices,’” which fail to provide an inventive step under Alice step two.
On appeal, a panel of the Federal Circuit first found the patents were directed to abstract ideas at Alice step one. The Court rejected Recentive’s argument that it did not claim generic machine learning technology because “the claims do not delineate steps through which the machine learning technology achieves an improvement.” Only performing “a task previously undertaken by humans with greater speed and efficiency,” does not make the technology patent eligible subject matter. The Court recognized that the specification had described the creation of network maps and event planning to be previously “manually created by humans” and reasoned that the claims simply “introduce machine learning techniques to the fields of event planning and creating network maps,” which improves “a task previously undertaken by humans with greater speed and efficiency.” The Court then proceeded to reason that “using machine learning to dynamically generate optimized maps and schedules based on real-time data and update them based on changing conditions” was “no more than claiming the abstract idea itself.” Thus, the Court held there was no inventive step.
The Court concluded that, although “[m]achine learning is a burgeoning and increasingly important field and may lead to patent-eligible improvements in technology[,] … patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.”
The opinion can be found here.
By Jeff Liao
ALSO THIS WEEK
Sage Products LLC v. Stewart, Appeal Nos. 2023-1603, -1604 (Fed. Cir. April 15, 2025)
In an appeal from inter partes reviews invalidating all challenged claims of two topical antiseptic patents owned by Sage Products, the Federal Circuit affirmed. On appeal, Sage contended that an artisan would not have understood prior art disclosing a “sterile” composition to have been “sterilized” within the meaning of its claims, including due to known mislabeling of the prior art product in the United States and differing applicable regulatory standards in the United Kingdom. The Federal Circuit rejected Sage’s arguments, finding that the Patent Trial and Appeal Board’s contrary finding was supported by substantial evidence. The Court also rejected Sage’s argument that the Board had erroneously found an anticipatory prior art reference to be enabled, holding that the Board was permitted to rely on evidence outside of the reference itself to show the state of the prior art and that the claimed subject matter was in the possession of the public.
The opinion can be found here.
By Jason A. Wrubleski
