Re.
Patent Application No. 2,144,068, CD 1339
This
decision was a review by the Patent Appeal Board of the Examiner’s final
rejection in respect of Canadian patent application No. 2,144,068 entitled
"FRAUD DETECTION USING PREDICTIVE MODELING". The primary issue on
review was whether the claims were directed to statutory subject matter.
The Patent
Appeal Board upheld the Examiner’s rejection and found that all the claims are
abstract and do not constitute patentable subject matter.
Claim
Construction
Background of the Invention
The subject
matter of the present application pertains generally to automated methods and
systems to detect fraudulent financial transactions, particularly in the use of
credit cards. [12] The application proposes to adapt a predictive model such as
a neural network in place of the parameter analysis model, to overcome the
prior art limitations in fraud detection. A neural network is a mathematical
model containing information representing the learned relationships among a
number of variables. [15]
Common General Knowledge
The Board
concluded that the general idea of using neural networks in financial
applications, including fraud detection, would have been part of the knowledge
of the skilled person. [23]
Practical Problem
The Board
stated that “the practical problem is how to adapt a predictive model (or
neural network) to improve credit card fraud detection over the previous
parameter based model.” [25]
Solution Proposed by the Application
The Board
found that “the solution proposed by the description is the provision of a
specific derived training data set based on past fraud-related variables,
transaction data and consumer profiles of historical spending, where the
training data set can be applied to a predictive model algorithm (neural
network) to process current transaction data to output a score value indicating
a likelihood of fraud.” [30]
Claims
Claim 1: A
computer-implemented process for identifying and determining fraudulent
transaction data in a computer-controlled transaction processing system
including predictive modeling means for receiving current transaction data,
processing the current transaction data, and outputting a plurality of output
values including a score value representing a likelihood of a fraudulent
transaction, comprising the steps of:
– prior to receiving the current transaction data for at least one
current transaction:
– generating a consumer profile for each of a plurality of consumers
from a plurality of past fraud-related variables and from consumer data, each
consumer profile describing historical spending patterns of a corresponding
consumer;
– the past fraud-related variables being derived by pre-processing past
transaction data, the past transaction data including values for a plurality of
transaction variables for a plurality of past transactions, the consumer data
including values for each consumer for a plurality of consumer variables;
– training the predictive modeling means with the consumer profiles and
with the past fraud-related variables to obtain a predictive model;
– storing the obtained predictive model in the computer;
– receiving current transaction data for a current transaction of a
consumer, receiving consumer data associated with the consumer;
– receiving the consumer profile associated with the consumer;
– pre-processing the obtained current transaction data, consumer data,
and consumer profile to derive current fraud-related variables for the current
transaction;
– determining the likelihood of fraud in the current transaction by
applying the current fraud-related variables to the predictive model; and
– outputting from the predictive modeling means an output signal
indicating the likelihood that the current transaction is fraudulent.
The Board
focused its analysis on whether or not the limitations of "a computer-implemented
process" and "a computer-controlled transaction processing
system" are essential features of the claimed solution. [32]
First, the
Board concluded that the feature of "a computer-controlled transaction
processing system" does not materially affect the working of the invention
(solution to the problem), and thus is not an essential element of the claim.
[35]
Turing next
to the term "computer-implemented process" as defined in the preamble
of claim 1, the Board acknowledged that “neural networks, as with any type of
algorithm or mathematical model, are typically executed using computers. Using
a computer is especially convenient, since such models tend to employ
mathematically-intensive calculations and use large amounts of data.” [36]
However, the Board noted that “needing a computer for practical convenience
(complicated calculations or large amounts of data) does not make the computer
essential for the working of an invention. Where a claim does not define a
solution to a computer ‘problem’, or overcome any technical problem in the
operation of the computer system, it points to the use of the computer as a
matter of convenience to perform calculations.” [36]
Based on
this reasoning, the Board concluded that “the computer implementation in the method
of claim 1 is not an essential element of the construed claim.” [38]
Next, the
Board considered a second independent claim, claim 12:
Claim 12: A
computer-controlled transaction processing system including predictive modeling
means for receiving current transaction data, processing the current
transaction data, and outputting a plurality of output values, including a
score value representing a likelihood of a fraudulent transaction, including:
– a model development component for developing a predictive model,
comprising:
– means for receiving past transaction data for a plurality of past
transactions, the past transaction data providing values for a plurality of
transaction variables;
– means for receiving consumer data for each of a plurality of
consumers, the consumer data providing values for a plurality of consumer
variables for each consumer;
– means for pre-processing the past transaction data to derive past
fraud related variables wherein at least some of the past fraud-related
variables are not present in the plurality of variables in the past transaction
data;
– means for generating a consumer profile for each individual consumer,
from the past fraud-related variables and the received consumer data, the
consumer profile describing historical spending patterns of the consumer;
– means for training the predictive model with the consumer profiles and
with the past fraud-related variables; and
– means for storing the trained predictive model in the computer; and
– a model application component, for applying the trained predictive
model, comprising:
– means for receiving current transaction data for a transaction of a
consumer;
– means for receiving consumer data associated with the consumer;
– means for receiving the consumer profile associated with the consumer;
– a current transaction data pre-processor, for pre-processing the
obtained current transaction data, consumer data, and consumer profile to
derive current fraud-related variables for the current transaction;
– means for determining the likelihood of fraud in the current
transaction by applying the current fraud-related variables to the predictive
model; and
– means for outputting from the predictive model an output signal
indicating the likelihood that the current transaction is fraudulent.
The Board
held that “[a]lthough defined in language of a system or machine claim, the
panel sees no material difference between the features of claim 12 and the
method steps of claim 1. Thus the panel considers that the essential features
of the claimed invention of claim 12 are equivalent to those of claim 1.
Accordingly, the computer and computerized components of claim 12 are not
essential to the invention.” [43]
In
addition, the Board held that these claim construction conclusions also applied
to the remaining, dependent claims.
Subject
Matter
The Board
drew an analogy to Schlumberger and stated that “claims 1 and 12 define
an attempt to patent a ‘method of collecting, recording, and analyzing’ data,
using a computer programmed according to a mathematical formula. The formula in
this case are the calculations, cost functions and weights of a neural network.
As in Schlumberger, the mere presence of a computer or other physical
tool in claim 1 or 12 does not render the otherwise abstract formula or
calculations in the neural network patentable.” [52]
The Board
considered the Applicant’s argument that "the complexities of programming
these steps into a computer are not so easily overlooked. Unlike Schlumberger,
the computer must be programmed with complex method steps to analyze, generate
and manipulate data which cannot be performed by hand in any practical manner,
if at all." [54] In rejecting this argument, the Board stated that
“[n]either the claims nor the description identify or elaborate on any required
complex programming or technical obstacles. Inventions by definition need not
be complex; complexity is not a test for statutory subject matter. In some
cases, overcoming a complex practical problem or disclosure of a complex
technical solution may inform the question of statutory subject matter in an
application. But the degree of complexity of an algorithm (e.g. a neural
network) does not automatically render a method claiming such an algorithm
statutory.” [55]
Next, the
Board considered the Applicant’s second argument that “if, following a
purposive construction, one considers the "actual invention", it is
physical (includes a computer) and causes a physical change/effect (output of
the signal to identify a fraudulent transaction). Accordingly, the applicant
contends that claim 1 (and claim 12) recites something that has a physical
existence, with physical elements essential to make the invention work, thus
further distinguishing the claims from Schlumberger.” [56]
The Board
also rejected the Applicant’s second argument, and stated that “the purposive
construction of the independent claims has found that the physical elements of
the computer or any other components are not essential to the invention.
Further…the output of the solution is a number, representing a fraud score
value, which by itself is abstract and has only intellectual meaning. We do not
find, as the applicant contends, that the score value as defined causes a
physical change or effect. Any further use or physical effect of the score
value is a feature beyond the solution of the present application, and is not
defined in either claim 1 or 12, or any dependent claim.” [57]
Commentary
This “post-Amazon one-click patent” case once more
highlights the endless controversy over computer implemented inventions. Again the authors note that while courts
presently have great difficulty finding a claim element to be non-essential
without evidence from the applicant/patentee Bell Helicopter, the
Patent Appeal Board has nonetheless found a claimed element in this application
to be non-essential, even while the applicant is clearly stipulating that the
element is essential. At the same time,
the authors are not aware of a basis in Canadian law that a claim must “…
define a solution to a computer ‘problem’, or overcome any technical problem in
the operation of the computer system …”
in order to qualify as statutory subject matter. Finally, the authors hope you enjoy
untangling the contradiction of, on the one hand of accepting the need for a
computer for practical convenience, but on the other hand declaring that the
computer was not essential for the working of the invention. Which is it
please?