Over five and a half years on from the Supreme Court's Alice vs. CLS Bank ruling, patentees, patent professionals, judges, and USPTO personnel are still wrestling with what it means for an invention to be eligible for patenting. This is especially true for software-related innovations. Despite the software-driven digital economy accounting for approximately 7% of the U.S. gross domestic product, software inventions can have a rough path to allowance and are likely to have their eligibility challenged post-issuance. The disconnect between software being one of the main sources of innovation in the last decade and its least-favored-nation status in patent law is largely due to Alice and its progeny.
In Alice, the Court set forth a two-part test to determine whether claims are directed to patent-eligible subject matter under 35 U.S.C. § 101. One must first decide whether the claim at hand involves a judicially-excluded law of nature, a natural phenomenon, or an abstract idea. If so, then one must further decide whether any element or combination of elements in the claim is sufficient to ensure that the claim amounts to significantly more than the judicial exclusion. But elements or combinations of elements that are well-understood, routine, and conventional will not lift the claim over the § 101 hurdle. While this inquiry is generally carried out as a matter of law, factual issues can come into play when determining whether something is well-understood, routine, and conventional.
How to apply this test in practice has been anything but clear. In response to complaints from applicants, patentees, and numerous other parties, the USPTO published its 2019 Revised Patent Subject Matter Eligibility Guidance in January and an update thereto in October.
Notably, the Guidance breaks the first part of the Alice test into two sub-steps. In the former, one determines whether the claim recites a judicial exception, such as an abstract idea. In the latter, one determines further "whether the recited judicial exception is integrated into a practical application of that exception." To focus the analysis, the Guidance indicates that all abstract ideas should fall into one of three categories: mathematical concepts, certain methods of organizing human activity, and mental processes. Think of these as the "Three M's" -- math, money, and mental steps.
In order to put its version of the Alice test into practice, the USPTO has been publishing a series of informative PTAB § 101 decisions. We discussed four of these as well as a fifth earlier in the year. Recently, a sixth, Ex Parte Linden, has been added to the list.
The Claimed Invention
The applicant's claimed invention is described as being directed to:
[S]tate-of-the-art speech recognition systems developed using end-to-end deep learning. In embodiments, the model architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, embodiments of the system do not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learn a function that is robust to such effects. A phoneme dictionary, nor even the concept of a "phoneme," is needed. Embodiments include a well-optimized recurrent neural network (RNN) training system that can use multiple GPUs, as well as a set of novel data synthesis techniques that allows for a large amount of varied data for training to be efficiently obtained. Embodiments of the system can also handle challenging noisy environments better than widely used, state-of-the-art commercial speech systems.
The PTAB deemed claim 11 representative. It recites:
A computer-implemented method for transcribing speech comprising:
receiving an input audio from a user;
normalizing the input audio to make a total power of the input audio consistent with a set of training samples used to train a trained neural network model;
generating a jitter set of audio files from the normalized input audio by translating the normalized input audio by one or more time values;
for each audio file from the jitter set of audio files, which includes the normalized input audio:
generating a set of spectrogram frames for each audio file;
inputting the audio file along with a context of spectrogram frames into a trained neural network;
obtaining predicted character probabilities outputs from the trained neural network; and
decoding a transcription of the input audio using the predicted character probabilities outputs from the trained neural network constrained by a language model that interprets a string of characters from the predicted character probabilities outputs as a word or words.
The Examiner's Rejection
During prosecution, the Examiner rejected claim 11 and claims depending therefrom under § 101 as being directed to patent-ineligible subject matter. Particularly, the Examiner viewed claim 11 as "using the predicted character probabilities (mathematical formula) to decode a transcription of the input audio into words or text data." The Examiner found that this was "similar to the court case Gottschalk v. Benson because the predicted character probabilities (mathematical formula or relationship) is used to convert or transcribe audio data into text data (words)." Thus, the Examiner concluded that the claim was directed to an invention falling into the mathematical concepts category.
The Examiner also characterized the claim as merely "normalizing the input audio data (manipulating data)," and "generating spectrogram frames based on each audio file (generating information sets based on prior information sets)," "using a mathematical formula to convert audio data into text data (Decoding)." The Examiner asserted that this implied that the claim was also abstract under the "certain methods of organizing human activity" and "mental process" categories because a "human can listen to an audio file and transcribe the audio data into text data which can all be done mentally."
Moreover, the Examiner found that the claim also fails under the second step of Alice. To that point, the Examiner wrote that the claimed invention "[d]oes not amount to significantly more since it is just decoding a transcription using a mathematical formula or relationship."
The PTAB's Decision
Applying the Revised Guidance, the PTAB began with the inquiry of whether the claim recites a judicial exception. Notably, the PTAB disagreed with the Examiner's contention that the claims involve either a mental process or a method of organizing human activity. The PTAB stated:
While transcription generally can be performed by a human, the claims here are directed to a specific implementation including the steps of normalizing an input file, generating a jitter set of audio files, generating a set of spectrogram frames, obtaining predicted character probabilities from a trained neural network and decoding a transcription of the input audio using the predicted character probability outputs. These are not steps that can practically be performed mentally. Nor do we see how the claimed invention recites organizing human activity.
The PTAB also concluded that the claims do not recite a mathematical concept, pointing out that while the specification discloses an algorithm, the claims do not recite the algorithm. Thus, under the Revised Guidance, no mathematical concept is recited. In particular, the PTAB referenced the USPTO's eligibility example 38, which explains that even if some claim limitations are based on mathematical concepts, the claim is not abstract unless at least one of those concepts are recited therein.
The PTAB went on to note that even if the claims were considered to recite one of the aforementioned categories of abstract ideas, the claims still "are not directed to an abstract idea because the alleged judicial exception is integrated into a practical application." Relying on statements made by the applicant, the PTAB observed that "the claims of the current application include specific features that were specifically designed to achieve an improved technological result" and "provide improvements to that technical field." As evidence of this, the PTAB looked to the specification, which describes a trained neural network used with a language model that "achieves higher performance than traditional methods on hard speech recognition tasks while also being much simpler."
Based on this analysis, the PTAB concluded that "the Examiner erred in determining that the claims are directed to an abstract idea."
Turning to the second part of Alice, the PTAB took further issue with the Examiner's reasoning. Particularly, the PTAB noted that "the Examiner concludes the claims do not include any additional elements that amounts to significantly more than a judicial exception but fails to provide sufficient factual support." Under Berkheimer v. HP, such a conclusory rationale for contending that a claim does not amount to significantly more cannot stand.
Thus, the PTAB reversed the Examiner's rejections of claim 11 and all of its dependent claims.
Analysis
This case is of note because it is one of the first in which the PTAB has confirmed that a machine learning invention can be non-abstract. While the USPTO's example 39 has suggested that one can claim a machine learning procedure without reciting any of the underlying mathematics, a mental process, or a method of organizing human activity, this decision affirms that is the case. The PTAB also seemed persuaded, based on statements made in the specification, that the claimed invention entails an improvement over previous techniques used to address the problem domain.
Of course, a district court or the Federal Circuit might agree or disagree. But, at the very least, the reasoning herein provides a roadmap for claim drafting and prosecution that may situate an invention to pass § 101 muster in the USPTO.
Ex Parte Linden (PTAB 2019)
Panel: Administrative Patent Judges Kumar, McKeown, and Shiang
Decision on Appeal by Administrative Patent Judge McKeown
