Event prediction and object recognition system

A computing device comprising:

• a processor;
• and
• a non-transitory computer-readable medium operably coupled to the processor, the computer-readable medium having computer-readable instructions stored thereon that, when executed by the processor, cause the computing device to
• read a dataset that includes a plurality of observation vectors;
• read a label set, wherein the label set defines permissible values for a target variable, wherein a value of the permissible values of the target variable is defined for a subset of the plurality of observation vectors;
• compute an affinity matrix using a kernel function and the plurality of observation vectors;
• compute a diagonal matrix by summing each row of the computed affinity matrix, wherein the sum of each row is stored in a diagonal of the row with zeroes in remaining positions of the row;
• compute a normalized distance matrix using the computed affinity matrix and the computed diagonal matrix;
• define a label matrix using the value of the target variable of each observation vector of the plurality of observation vectors
• (a) initialize a classification matrix as the defined label matrix;
• (b) compute a converged classification matrix, wherein the converged classification matrix defines a label probability for each permissible value defined in the label set for each observation vector of the plurality of observation vectors, wherein the classification matrix is converged using F(t+1)=αSF(t)+(1−α)Y, where F(t+1) is a next classification matrix, α is a relative weighting value, S is the normalized distance matrix defined as S=D−1/2WD−1/2, where W is the computed affinity matrix and D is the computed diagonal matrix, F(t) is the classification matrix, Y is the label matrix defined as Yik=1 if xi is labeled as yi=k;
• otherwise, Yik=0, where xi is an observation vector of the plurality of observation vectors, i=1, . . . , n, n is a number of vectors of the plurality of observation vectors, k=1, . . . , c, and c is a number of permissible values of the label set, and t is an iteration number, wherein the classification matrix is converged when a second predefined number of iterations of computations of F(t+1)=αSF(t)+(1−α)Y is complete;
• (c) for each observation vector, update the value of the target variable based on a maximum label probability value identified from the converged classification matrix;
• a predefined number of times,
• compute a distance vector that defines a distance value between each pair of the plurality of observation vectors using a distance function applied to only the converged classification matrix;
• select a number of observation vectors from the dataset that have minimum values for the distance value;
• request that a user provide a label for each of the selected observation vectors;
• receive a response to the request from the user for each of the selected observation vectors;
• update the value of the target variable for each of the selected observation vectors with the received response;
• and
• repeat operations (a) to (c);
• and
• after the predefined number of times, output the value of the target variable for each observation vector of the plurality of observation vectors to a second dataset.

A method of predicting occurrence of an event or classifying an object using semi-supervised data to label unlabeled data in a dataset, the method comprising:

• reading, by a computing device, a dataset that includes a plurality of observation vectors;
• reading, by the computing device, a label set, wherein the label set defines permissible values for a target variable, wherein a value of the permissible values of the target variable is defined for a subset of the plurality of observation vectors;
• computing, by the computing device, an affinity matrix using a kernel function and the plurality of observation vectors;
• computing, by the computing device, a diagonal matrix by summing each row of the computed affinity matrix, wherein the sum of each row is stored in a diagonal of the row with zeroes in remaining positions of the row;
• computing, by the computing device, a normalized distance matrix using the computed affinity matrix and the computed diagonal matrix;
• defining, by the computing device, a label matrix using the value of the target variable of each observation vector of the plurality of observation vectors
• (a) initializing, by the computing device, a classification matrix as the defined label matrix;
• (b) computing, by the computing device, a converged classification matrix, wherein the converged classification matrix defines a label probability for each permissible value defined in the label set for each observation vector of the plurality of observation vectors, wherein the classification matrix is converged using F(t+1)=αSF(t)+(1−α)Y, where F(t+1) is a next classification matrix, α is a relative weighting value, S is the normalized distance matrix defined as S=D−1/2WD−1/2, where W is the computed affinity matrix and D is the computed diagonal matrix, F(t) is the classification matrix, Y is the label matrix defined as Yik=1 if xi is labeled as yi=k;
• otherwise, Yik=0, where xi is an observation vector of the plurality of observation vectors, i=1, . . . , n, n is a number of vectors of the plurality of observation vectors, k=1, . . . , c, and c is a number of permissible values of the label set, and t is an iteration number, wherein the classification matrix is converged when a second predefined number of iterations of computations of F(t+1)=αSF(t)+(1−α)Y is complete;
• (c) for each observation vector, updating, by the computing device, the value of the target variable based on a maximum label probability value identified from the converged classification matrix;
• a predefined number of times,
• computing, by the computing device, a distance vector that defines a distance value between each pair of the plurality of observation vectors using a distance function applied to only the converged classification matrix;
• selecting, by the computing device, a number of observation vectors from the dataset that have minimum values for the distance value;
• requesting, by the computing device, that a user provide a label for each of the selected observation vectors;
• receiving, by the computing device, a response to the request from the user for each of the selected observation vectors;
• updating, by the computing device, the value of the target variable for each of the selected observation vectors with the received response;
• and
• repeating, by the computing device, operations (a) to (c);
• and
• after the predefined number of times, outputting, by the computing device, the value of the target variable for each observation vector of the plurality of observation vectors to a second dataset.