Heuristic digital processor using non-linear transformation

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced,
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit, wherein said non-linear transforming means comprises a digital arithmetic unit.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced,
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit, wherein the processing means comprises programmed processing devices for performing calculation operations in parallel with one another.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced.
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, wherein the processing means comprises digital electronic signal processing means for performing calculations in floating point arithmetic, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced, wherein the transforming means is a digital arithmetic unit for performing calculations with fixed point arithmetic,
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced,
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit, wherein the transforming means and the processing means incorporate digital electronic signal processing devices controlled by clock signals.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced,
• processing means for weighting and combining training .phi. vector elements, wherein the processing means and for producing a training fit to a set of training answers comprises digital electronic signal processing devices for storing processing results for output after a delay, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit.

An heuristic processor comprised of:

• non-linear transforming means for producing a respective training .phi. vector from each member of a training data set on the basis of a set of centers, each training data set member having a displacement from each of said centers, where a norm of the displacement is calculable from each displacement, and each element of a training .phi. vector consisting of a non-linear transformation of the norm of the displacement of a respective center from the training data set member from which said training .phi. vector is produced,
• processing means for weighting and combining training .phi. vector elements and for producing a training fit to a set of training answers, and
• means for generating result estimate values, said generating means comprising means for producing a respective test .phi. vector from each member of a set of test data, each test data set member having a displacement from each of said centers, where a norm of said test data set member displacement is calculable from each test data set member displacement and each element of a test .phi. vector consisting of said non-linear transformation of said norm of said test data set member displacement, each of said estimate values consisting of a combination of the elements of a respective test .phi. vector and each said combination being at least equivalent to a summation of vector elements weighted in accordance with the training fit, wherein the processing means is a systolic array of processing cells for implementing a rotation algorithm to provide triangularization of a .PHI. matrix of .phi. vector rows and least squares fitting to the training answers set, the algorithm involving computation and application of rotation parameters and storage of updated triangularized matrix elements by the processing cells, and wherein the systolic array has a first row of processing cells arranged to receive composite vectors each comprising a respective .phi. vector and a respective training answer, each first row cell being arranged for input of a respective element of each composite vector.