EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

PROCESSING APPLICATIONS

In 1996 the Patent and Trademark Office (PTO) issued the Examination Guidelines for Computer-Related Inventions (Guidelines). Following issuance of the Guidelines, the PTO prepared and published training materials illustrating how to apply them.

Recently the PTO determined that additional training materials were needed to address how to apply the Guidelines in the areas of business, artificial intelligence and mathematical processing applications. Each of these three areas has shown a high g rowth rate and increased examining complexity. Thus, the following five examples were prepared to illustrate how to apply the Guidelines in these areas.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

The disclosure relates to a computerized method of evaluating investment risk factors between a plurality of mutual funds and optimizing an investment value to be distributed among the funds. The specification recites a general purpose digital comp uter suitably programmed to accomplish the method. No specific hardware is disclosed and no computer listing is provided, only high level flow diagrams and descriptions of the desired functionality. The specification contains numerous formulas which are used to calculate risk factors, distribution amounts, time periods, performance data on past transactions etc. These are described in great detail and it is noted that it is well within the skill of a programmer in the art to create the programs to acco mplish the required calculations. The specification indicates that communication between the investor, and a broker/fund manager takes place through the system. No specific form of communication is disclosed but the specification implies that such commu nication is part of the computer system.

The method involves storing in the computer memory data representing various mutual funds (identifiers) as well as risk ranking factors for each fund. Individual investor profiles are established for a particular time frame and these data are also sto red in the computer. The method is implemented when an investor specifies a dollar amount to be invested. The computer then calculates the optimal disbursement of the allocation between various funds to meet the investor profile previously established.< /P>

The disclosure presents several embodiments which act to: (1) merely advise the investor on possible investment strategies, (2) prepare a report of investment strategies to be incorporated into a monthly investor account summary, or (3) control an aut omated scheme (computer controlled) to buy and sell shares of mutual funds in order to invest according to the optimized profile.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Claim 1

A computerized method of allocating funds for a mutual fund among a plurality of funds in a group, comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table Notes for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Claim 2

The method of claim 1, further including the step of displaying the optimum account allocation on an investor monthly account summary report to an investor or broker.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table Notes for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Claim 3

The method of claim 1, further including the step of transferring funds between the mutual funds in the group according to the optimum account allocation.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Mutual Fund

Table Notes for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

The specification discloses a method of performing matrix multiplication using a general purpose computer. No specific computer hardware or software programs are disclosed. The specification recites specific algorithms for manipulating matrices i ncluding the multiplication of two matrices together. A flow chart showing the steps involved in creating the rows and columns of the matrix and the multiplication of the terms is provided. The terms of the matrix are disclosed as representing vectors. The vectors could represent data collected from real world objects or they could be abstractions of non physical systems.

The method consists of creating two matrices having terms defined by disclosed mathematical relationships such as being non-zero, and related to a prime number or a factorial of a prime number. After creating the two matrices they are combined into on e matrix by interleaving rows and columns until a prescribed mathematical relationship exists. A multiplication of the matrixes then takes place whereby an output result is determined which defines the value(s) for some unknown quantity.

The disclosure provides several examples of possible uses for the method which include simulation of space craft flight paths. The specification mentions that if this method were incorporated into the control environment of a space craft the pilot cou ld use the method to optimize flight paths. No details of how this would be done are recited in the specification but the disclosure complies with the requirements of 35 U.S.C. § 112. The disclosure states that the invention is not limited to the sp ace craft environment.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Claim 1

A processing system for performing a plurality of matrix manipulations comprising:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table Notes for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Claim 2

A processing system for modeling space craft thruster operation by performing a plurality of matrix manipulations of terms representing thrust vectors comprising:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table Notes for Claim 2

THE REMAINDER OF THE EXAMINATION MUST BE COMPLETED.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Claim 3

A processing system for modeling space craft thruster operation to aid pilots in control of the vehicle by performing a plurality of matrix manipulations of terms representing thrust vectors comprising:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Matrix

Table Notes for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

The specification discloses a method of training a neural network node using a general purpose computer. The general purpose computer contains a CPU and a math coprocessor. The computer has a standard operating system and configuration for memor y having a number of interconnected memory cells each working together. The method consists of a sequence of functions being carried out in a specific order to achieve the functionality of training this specialized network to perform a wide range of vari ed functions. The method of training a neural network node contains a number of basic steps. The first is a step of providing an initial set of target points in the model space. After the set of target points is set then an estimate of the probability density function (PDF) on the model space at each node in the model space is generated. After the PDF is generated for each node then a second set of target points in said model space is determined. The second set of target points are individually or co mbinatorially evaluated using the probability density function PDF.

For optimum training within a system a threshold value must be selected for the desired functionality. The threshold value is determined to be less than PDF(i) , where i is the ith target point for each of the second set of target points. Using this threshold value, a first training set of target points for the model space is computed using the N.N.S.(Neural Network Standard determined by National Institute for Standards and Technology in 1995) where the input value is selected and the output value i s the PDF(Input value) where PDF(Input value) less than the threshold value. Using this threshold value, a second training set of target points for the model space is computed using the N.N.S.(Neural Network Standard determined by National Institute for Standards and Technology in 1995) where the input value is selected and the output value is the PDF(Input value) where PDF(Input value) greater than said threshold value. Once the first and second set of training sets are determined to meet the criteria set forth above, these sets now contain the desired characteristics to appropriately train the neural network for the desired functionality.

There are a wide range of functions which may be carried out by the ultimate end user of the neural network. The function will dictate the criteria upon which the network specifications must be established. The process will vary upon the selection o f the criteria. The disclosed methodology is the basic framework from which most functionalities may be established from an appropriate training set.

The neural networks to be trained may be either based in a hardware based system which is adapted or it may similarly be based upon a general purpose computer to carry out the desired functionality as a neural network. The training may be done either by a technician, by automated system or by a programed system in the general purpose computer.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Claim 1

A computerized method of training a neural network node comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table Notes for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Claim 2

A computerized method of training a neural network node comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table Notes for Claim 2

Example: Neural Network

Claim 3

A method of determining the state of a neural network comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Neural Network

Table Notes for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

The specification discloses a method of performing triple precision arithmetic in a general purpose computer. The general purpose computer contains a CPU and a math coprocessor. The computer has a standard operating system and configuration for memor y having a number of interconnected memory cells each working together. The method is a series of steps being performed on the general purpose computer to perform the arithmetic addition of at least two triple precision values into a resultant triple qua ntity. Prior art methods have only consisted of double precision computations. Many of the complex chemical analysis computations which previously were performed on supercomputers, may now be carried out by desktop computer with the addition of a dedica ted subsystem for performing simultaneous computations while the computer carries out other data manipulations. An embodiment of such an implementation is set forth in the detailed description. The dedicated processing of the mathematical computations al ong with the added precision of the triple precision mathematics allow the desktop computer to be a stand alone design and perform as an analysis workstation for a wide array of applications in mathematics and the sciences which is unrivaled by systems on the market today.

The method is composed of a series of steps which must be performed in a specified order to achieve the triple precision computation. The first step is converting a first triple precision quantity into a first pair of overlapping double precision quan tities and storing two copies of the first overlapping double precision quantities into memory cells. The conversion is performed by a known method of dividing the numerical representation of the value into two equal number of bytes which are each equal to the maximum word size for a double precision value. The second step is converting said second triple precision quantity into a second pair of overlapping double precision quantities and storing two copies of the second overlapping double precision qua ntities into memory cells. The conversion is performed in the same manner as recited in the first step above. The third step of the method generates a first intermediate value by performing at least one double precision arithmetic operation on a first c opy of each said first and second double precision quantities. Depending upon the level of accuracy desired in the computation, a number of computations may be made and a statistical mean value may be used in the determination of the actual value of the result. The fourth step is generating a second intermediate value by performing the inverse operation of said at least one double precision arithmetic operation on said second copy of each said first and second double precision quantities. Again, dependi ng upon the level of accuracy desired in the computation, a number of computations may be made and a statistical mean value may be used in the determination of the actual value of the result.

The statistical evaluation of each of the intermediate values may be made individually to generate the intermediate values which are then combined combinatorially to generate the result of the triple precision arithmetic operation. Alternatively, the individual intermediate values may by combinatorially combined to generate the triple precision arithmetic value. The statistical manipulation of the values would have been within the level of skill of those in the art and various computer programs could be generated to implement the functions set forth in the high level flow diagrams attached in Appendix 1to the specification. Figures 1A and 1B show one of the preferred hardware embodiments for performing the same arithmetic operations as described in the flow diagrams. Due to the limitations on clock speed, the hardware embodiment has been the more efficient but more costly in implementation of the method on a wide range of computers. The software embodiment has proven to be the more flexible and mo re implementable embodiment of the invention than the hardware embodiment where the speed is not critical. As higher clock speeds become available the hardware and software embodiments will function at comparable speeds and efficiency.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Claim 1

A method of performing triple precision arithmetic in a computer having a plurality of memory cells comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table Notes for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Claim 2

A triple precision arithmetic logic unit in a computer having a plurality of memory cells comprising:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table Notes for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Claim 3

A triple precision arithmetic logic unit as claimed in claim 2, wherein the logic unit is embedded in a single monolithic PAL.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Triple Precision Arithmetic

Table Notes for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

The invention is a method of determining whether to extend real estate services to a potential customer. The method is performed utilizing a general purpose computer system configured for that purpose, i.e., specific inputs for receiving data, ALUs, outputs, etc. for implementing the method.

The method includes a series of steps to be performed on a computer system for providing real time indications of whether to extend real estate services such as insurance, second mortgages, lines of credit, etc. based upon the potential customer’s owne rship of specified securities. The computer system receives data relating to the value of the specified securities, market variations/changes with respect to those securities, and a potential range of acceptable future values for those securities. Employ ing this data, the computer system determines the probable value of those securities at a time in the future to assess the risk of extending a home mortgage service or other real estate related service to the potential customer who is the owner of the sec urities. The assessment is made by comparing a determined level of risk for extending the service with a threshold value for that risk. The outcome of the evaluation and the resulting decision on the real estate related service are conveyed to the poten tial customer.

The disclosed invention includes a preferred embodiment in specific hardware/software but also includes high level flow charts that could be used to implement the method in "any and every" product. The notification to the potential customer is disclos ed as including the preparation of a "form letter" of acceptance/rejection; but further includes a general statement that "any other appropriate means" could be used. The "form letter" can be prepared by the computer system-printer output or prepared by a person.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Claim 1

In a system for real time determination of a market indicator for securities which mature within a set time, the system comprising:

the method of determining the level of risk in extending a real estate service comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table Notes for Claim 1

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Claim 2

In a system for real time determination of a market indicator for securities which mature within a set time, the system comprising:

the method of determining whether to extend a real estate service comprising the steps of:

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table Notes for Claim 2

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Claim 3

In the method of claim 2, the further step of notifying a potential buyer of the decision on whether to extend the related real estate service.

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table for Claim 3

EXAMINATION GUIDELINES FOR COMPUTER-RELATED INVENTIONS

Example: Market Securities

Table Notes for Claim 3