Method, system, and computer program product for representing object relationships in a multidimensional space
First Claim
1. A method of mapping a set of input patterns to an m-dimensional space, wherein each pattern represents a compound selected from a database of compounds, the method comprising the steps of:
- (a) selecting k patterns from said set of input patterns to form a subset of patterns {pi, i=1, . . . , k};
(b) determining at least some pairwise relationships between at least some of the patterns in said subset of patterns {pi}(c) mapping the patterns {pi} into a set of images in an m-dimensional space {pi→
yi, i=1, 2, . . . k, yi∈
Rm) so that at least some of the pairwise distances between at least some of the images {yi} are representative of the relationships of the respective patterns {pi};
(d) determining a set of n attributes for each pattern in said subset of patterns {pi} {xi, i=1, 2, . . . k, xi∈
Rn};
(e) forming a training set T={(xi, yi), i=1, 2, . . . k};
(f) using a supervised machine learning technique to determine a mapping function based on the training set T; and
(g) using said mapping function determined in step (f) to map additional patterns.
2 Assignments
0 Petitions
Accused Products
Abstract
A method, system, and computer product is presented for mapping a set of patterns into an m-dimensional space so as to preserve relationships that may exist between these patterns. A subset of the input patterns is chosen and mapped into the m-dimensional space using an iterative nonlinear mapping process based on subset refinements. A set of n attributes are determined for each pattern, and one or more neural networks or other supervised machine learning techniques are then trained in accordance with the mapping produced by the iterative process. Additional input patterns not in the subset are mapped into the m-dimensional space by determining their n input attributes and using the neural networks in a feed-forward (prediction) mode.
-
Citations
27 Claims
-
1. A method of mapping a set of input patterns to an m-dimensional space, wherein each pattern represents a compound selected from a database of compounds, the method comprising the steps of:
-
(a) selecting k patterns from said set of input patterns to form a subset of patterns {pi, i=1, . . . , k}; (b) determining at least some pairwise relationships between at least some of the patterns in said subset of patterns {pi} (c) mapping the patterns {pi} into a set of images in an m-dimensional space {pi→
yi, i=1, 2, . . . k, yi∈
Rm) so that at least some of the pairwise distances between at least some of the images {yi} are representative of the relationships of the respective patterns {pi};(d) determining a set of n attributes for each pattern in said subset of patterns {pi} {xi, i=1, 2, . . . k, xi∈
Rn};(e) forming a training set T={(xi, yi), i=1, 2, . . . k}; (f) using a supervised machine learning technique to determine a mapping function based on the training set T; and (g) using said mapping function determined in step (f) to map additional patterns. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22)
-
-
23. A computer program product comprising a computer useable medium having computer program logic recorded thereon for enabling a processor to obtain pairwise relationship data about the selected pair of patterns, wherein each pattern represents a compound selected from a database of compounds, said computer program logic comprising:
-
a selecting procedure that enables the processor to select a plurality of patterns from a database for similarity; a transmitting procedure that enables a processor to transmit selected patterns to a remote computer for similarity comparison; and a receiving procedure that enables the processor to receiving similarity data about patterns transmitted to the remote computer. - View Dependent Claims (24, 25)
-
-
26. A system for mapping a set of input patterns to an m-dimensional space, wherein each pattern represents a compound selected from a database of compounds, the system comprising:
-
means for selecting k patterns from said set of input patterns to form a subset of patterns {pi, i=1, . . . , k}; means for determining at least some pairwise relationships between at least some of the patterns in said subset of patterns {pi}; means for mapping the patterns {pi} into a set of images in an m-dimensional space {pi→
yi, i=1, 2, . . . k, yi ∈
Rm) so that at least some of the pairwise distances between at least some of the images {yi} are representative of the relationships of the respective patterns {pi};means for determining a set of n attributes for each pattern in said subset of patterns {pi}, {xi, i=1, 2, . . . k, xi ∈
Rn};means for forming a training set T={xi, yii=1, 2, . . . k}; means for using a supervised machine learning technique to determine a mapping function based on the training set T; and means for using said mapping function determined in step (f) to map additional patterns.
-
-
27. A computer program product comprising a computer useable medium having computer program logic recorded thereon for enabling a processor to obtain map a set of input patters to an m-dimensional space, wherein each pattern represents a compound selected from a database of compounds, said computer program logic comprising:
-
a procedure that selects k patterns from said set of input patterns to form a subset of patterns {pi, i=1, . . . , k}; a procedure that determines at least some pairwise relationships between at least some of the patterns in said subset of patterns {pi}; a procedure that maps the patterns {pi} into a set of images in an m-dimensional space {pi→
yi, i=1, 2, . . . k, yi ∈
Rm) so that at least some of the pairwise distances between at least some of the images are representative of the relationships of the respective patterns {pi};a procedure that determines a set of n attributes for each pattern in said subset of patterns {pi}, {xi, i=1, 2, . . . k, xi ∈
Rn};a procedure that forms a training set T={xi, yi), i=1, 2, . . . k }; a procedure that uses a supervised machine learning technique to determine a mapping function based on the training set T; and a procedure that uses said mapping function determined in step (f) to map additional patterns.
-
Specification