AI and blockchain: a practical guide to overcoming 101 rejections

25-05-2023

Alexander Taousakis and Bradford Fritz

AI and blockchain: a practical guide to overcoming 101 rejections

Yurchanka Siarhei / Shutterstock.com

Focusing on a practical use for computer-based inventions is one tactic to ensure patent applications don’t short circuit, say Alexander Taousakis and Bradford Fritz of Birch, Stewart, Kolasch & Birch.

Overcoming a 101 rejection can be challenging, especially for computer-based patent applications such as inventions directed to artificial intelligence (AI) and blockchain technology (think non-fungible tokens or cryptocurrency). However, focusing on reciting a “practical application” in the claim and arguing how the invention is an improvement over the existing technology can be a good tactic when facing a 101 rejection.

While there are many ways to successfully argue a claim is statutory, MPEP 2106.05 presents two different tests for determining whether a claim is statutory under 35 U.S.C. § 101:

1) whether the claim is tied to a particular machine that is integral to the claim or involves a process that includes a transformation and reduction of the article to a different state or thing, or

2) whether the claim is directed to an improvement to other technology or technical field, as discussed in MPEP §§ 2106.04(d)(1) and 2106.05(a) (eg, Revised Step 2A, prong two of the streamlined analysis).

Option 1) involves reciting enough hardware elements in the claim, and arguing these features are more than generic computer components and are significantly more than the abstract idea. This option is often weak, especially where the claim mainly recites just a processor or controller configured with an algorithm. Also, this determination of whether enough hardware elements are present in the claim can be rather subjective.

Therefore, the focus should be directed to Option

2). This approach centres around reciting a practical application within the claim, such as the end result of a real-world use case scenario. For example, claims rejected under Section 101 for reciting an abstract idea often stop short and end with a final determination, rather than using that determination to perform a real-world action. This position can be strengthened by explaining how the claim provides a technical solution to an existing problem, including identifying the problem/solution discussed in the specification, referencing the pertinent paragraphs, and clarifying how it relates to the claimed features.

The improvement is not limited to the functioning of a computer (eg, faster processing speed, etc.). Instead, as outlined in MPEP §§ 2106.04(d)(1) and 2106.05(a), the improvement can relate to “other technology or technical field”. An improvement to “other technology or technical field” is rather broad and often easy to satisfy, which can be brought to the examiner’s attention.

Below are two examples that may aid in either responding to an Office Action or persuading the Patent Trial and Appeal Board (PTAB) to reverse a 101 rejection.

Blockchain example

Ex parte Steven Charles Davis is a PTAB case for US Application No. 14/719,030 (US 10,963,881, herein the ’881 Patent), entitled “Method and System for Fraud Control of Blockchain-based Transactions”.

"An improvement to ‘other technology or technical field’ is rather broad and often easy to satisfy, which can be brought to the examiner’s attention."Claims 1-16 were rejected under 35 U.S.C. § 101 for being directed to an abstract idea. The invention processes blockchain transactions using transaction messaging and traditional payment networks (’881 Patent, column 8, lines 41-47). By using traditional payment networks, the transaction is more secure and consumers can engage in blockchain transactions “without being in constant possession of a computing device that stores their private keys” (’881 Patent, column 8, lines 46-54).

Under Step 2A, Prong 1 of the US Patent and Trademark Office (USPTO), 2019 Revised Patent Subject Matter Eligibility Guidance, the examiner concluded that the claims are directed to an abstract idea of a mental process coupled with the use of a conventional/generic computer, and not to an improvement in server functionality or improvements to a technological process.

The appellant argued the transaction messages are processed at nanosecond speed (due to the frequency of transactions) and “of sufficient data size and complexity to not be understood by human mental work”, which necessarily requires specialised computer systems, as opposed to the conventional computer contemplated by the examiner.

Under Step 2A, Prong 2, the PTAB concluded the claims recite a practical application by providing the security of standard payment processing systems and the privacy of blockchain payment transactions to verify a digital signature.

Further, the PTAB found the claimed use of “an account database of a computer system”, “a receiver of the computer system”, “a payment network”, “a processor”, and “a blockchain network” represents an ordered combination “that achieves a technological improvement”.

In other words, the appellant established the claimed invention could not be interpreted as a mental process by proving that the claimed steps require a specialised computer that processes the transaction messages at “speeds that have to be measured in nanoseconds for network reliability and due to the overwhelming number of transactions processed each day” (Reply Br. 2-4). Further, the appellant established the claim includes an ordered combination that recites an inventive concept that is an unconventional solution to the technical problem of fraud and theft (‘881 Patent column 2, lines 15-20).

The main takeaway from this case is to present arguments directed to any improvement the claimed invention has over known solutions and, if possible, argue a specialised computer system would be required to perform the claimed steps.

AI example

US 10,268,950 (herein the ‘950 Patent) is directed to image processing using machine learning (ie, a convolutional neural network (CNN)). A computer is automatically trained using a pool of face images and non-face images until a certain success rate is achieved (‘950 Patent Abstract and column 1, line 50—column 2, line 12).

In the Final Rejection dated July 13, 2018, the claims were rejected under 35 U.S.C. § 101 as being directed to “the abstract idea of processing/analysing gathering data for training and validation of a convolutional neural network”, which is “similar to other ideas found to be abstract by various courts, such as collecting information, analysing it, and displaying certain results of the collection and analysis”, citing Electric Power Group.

The claim did recite some training of the CNN. Generating a trained neural network can sometimes satisfy Section 101. However, the claim lacked an application of the CNN (eg, transforming the abstract idea into a practical application).

To address 35 U.S.C. § 101 rejection, the claim was amended to recite “incorporating the trained CNN in a face detection system; and using the face detection system with [the] trained CNN to detect faces in images”. By tying the claim to a face detection system and detecting faces in images with the trained CNN, the claim integrates the alleged abstract idea into a practical application. For example, rather than the end result of the claim being directing to optimising the neural network, the amended claim applied the trained neural network in a face detection system to detect faces in images.

That is, by reciting a face detection system and its use, the claim applied “the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolise the exception” MPEP 2106.05(e).

MPEP 2106.04(a)(1) includes a similar situation, involving the following:

a method of training a neural network for facial detection comprising:

collecting a set of digital facial images;

applying one or more transformations to the digital images;

creating a first training set including the modified set of digital facial images;

training the neural network in a first stage using the first training set;

creating a second training set including digital non-facial images that are incorrectly detected as facial images in the first stage of training; and training the neural network in a second stage using the second training set.

In this example, reciting a complex type of training of a neural network, involving multiple stages and different training sets, is considered enough to be statutory under 35 U.S.C § 101.

"The appellant established the claimed invention could not be interpreted as a mental process by proving that the claimed steps require a specialised computer."

These steps solve the inability of prior art methods to “robustly detect human faces in images where there are shifts, distortions, and variations in scale and rotation of the face pattern in the image” (Subject Matter Eligibility Example 39). In the analysis, it is determined that the claims do not recite a mental process and further, that they “are not practically performed in the human mind” (Id.).

If faced with a similar situation, such as a claim directed to training a neural network, initially, cite Example 39 and argue the claims do not recite a mental process, the steps (if applicable) cannot be “practically performed in the human mind”, for example, due to the sheer number of calculations or their complexity, and/or the claim solves a problem and thus involves a technological solution.

As a back-up position, the claims could be amended to use this trained neural network to provide an improvement to a particular technology, such as using the trained neural network to do something useful, as in the above AI example. Emphasise the disclosed improvement, and it may be helpful to conduct an interview with the examiner.

The above examples show how reciting a real-world use or action in the claim that is directed to solving a problem in the existing technology can help integrate the alleged abstract idea into a practical application and overcome a 101 rejection.

Alexander Taousakis is an associate at BSKB who formerly served as a patent examiner at the USPTO. He can be contacted at: ataousakis@bskb.com

Bradford Fritz is an associate at BSKB and former patent examiner at the USPTO. He can be contacted at: bfritz@bskb.com

Birch, Stewart, Kolasch & Birch, AI, blockchain, inventions, patent applications, technology, NFTs, cryptocurrency, USPTO, PTAB, neural network

WIPR