Sas Institute Inc.

Xu Chen

Brett Alan Wujek

Jorge Manuel Gomes Da Silva

Saratendu Sethi

Yingjian Wang

Tao Wang

Xinmin Wu

A computing device classifies unclassified observations. A first batch of unclassified observation vectors and a first batch of classified observation vectors are selected. A prior regularization error value and a decoder reconstruction error value are computed. A first batch of noise observation vectors is generated. An evidence lower bound (ELBO) value is computed. A gradient of an encoder neural network model is computed, and the ELBO value is updated. A decoder neural network model and an encoder neural network model are updated. The decoder neural network model is trained. The target variable value is determined for each observation vector of the unclassified observation vectors based on an output of the trained decoder neural network model. The target variable value is output.

Machine learning classification system

Unclassified observations are classified. Similarity values are computed for each unclassified observation and for each target variable value. A confidence value is computed for each unclassified observation using the similarity values. A high-confidence threshold value and a low-confidence threshold value are computed from the confidence values. For each observation, when the confidence value is greater than the high-confidence threshold value, the observation is added to a training dataset and, when the confidence value is greater than the low-confidence threshold value and less than the high-confidence threshold value, the observation is added to the training dataset based on a comparison between a random value drawn from a uniform distribution and an inclusion percentage value. A classification model is trained with the training dataset and classified observations. The trained classification model is executed with the unclassified observations to determine a label assignment.

Semi-supervised classification system

Data is classified using semi-supervised data. Sparse coefficients are computed using a decomposition of a Laplacian matrix. (B) Updated parameter values are computed for a dimensionality reduction method using the sparse coefficients, the Laplacian matrix, and a plurality of observation vectors. The updated parameter values include a robust estimator of a decomposition matrix determined from the decomposition of the Laplacian matrix. (B) is repeated until a convergence parameter value indicates the updated parameter values for the dimensionality reduction method have converged. A classification matrix is defined using the sparse coefficients and the robust estimator of the decomposition of the Laplacian matrix. The target variable value is determined for each observation vector based on the classification matrix. The target variable value is output for each observation vector of the plurality of unclassified observation vectors and is defined to represent a label for a respective unclassified observation vector.

Distributable event prediction and machine learning recognition system

A computing device predicts occurrence of an event or classifies an object using distributed unlabeled data. A Laplacian matrix is computed using a kernel function. A predefined number of eigenvectors is selected from a decomposed Laplacian matrix to define a decomposition matrix. A gradient value is computed as a function of the defined decomposition matrix, a plurality of sparse coefficients, and a label matrix, a value of each coefficient of the plurality of sparse coefficients is updated based on the computed gradient value, and the computations are repeated until a convergence parameter value indicates the plurality of sparse coefficients have converged. A classification matrix is defined using the plurality of sparse coefficients to determine the target variable value for each observation vector of the plurality of unclassified observation vectors. The target variable value for each observation vector of the plurality of unclassified observation vectors is output.

A computing device classifies unclassified observations. A first batch of noise observations is generated. (A) A first batch of unclassified observations is selected. (B) A first batch of classified observations is selected. (C) A discriminator neural network model trained to classify unclassified observations and noise observations is updated with observations that include the first batch of unclassified observations, the first batch of classified observations, and the first batch of noise observations. (D) A discriminator loss value is computed that includes an adversarial loss term computed using a predefined transition matrix. (E) A second batch of unclassified observations is selected. (F) A second batch of noise observations is generated. (G) A generator neural network model trained to generate a fake observation vector for the second batch of noise observations is updated with the second batch of unclassified observations and the second batch of noise observations. (H) (A) to (G) is repeated.

Data is classified using semi-supervised data. A decomposition is performed to define a first decomposition matrix that includes first eigenvectors of a weight matrix, a second decomposition matrix that includes second eigenvectors of a transpose of the weight matrix, and a diagonal matrix that includes eigenvalues of the first eigenvectors. Eigenvectors are selected from the first eigenvectors to define a reduced decomposition matrix. A linear transformation matrix is computed as a function of the first decomposition matrix, the reduced decomposition matrix, the diagonal matrix, and a penalty matrix. When a rank of the linear transformation matrix is less than a number of rows of the penalty matrix, a classification matrix is computed by updating a gradient of a cost function. When the rank of the linear transformation matrix is equal to the number of rows of the penalty matrix, the classification matrix is computed using a dual formulation.

Data is classified using semi-supervised data. A weight matrix is computed using a kernel function applied to observation vectors. A decomposition of the computed weight matrix is performed. A predefined number of eigenvectors is selected from the decomposed weight matrix to define a decomposition matrix. (A) A gradient value is computed as a function of the defined decomposition matrix, sparse coefficients, and a label vector. (B) A value of each coefficient of the sparse coefficients is updated based on the gradient value. (A) and (B) are repeated until a convergence parameter value indicates the sparse coefficients have converged. A classification matrix is defined using the converged sparse coefficients. The target variable value is determined and output for each observation vector based on the defined classification matrix to update the label vector and defined to represent the label for a respective unclassified observation vector.

Data is classified using corrected semi-supervised data. Cluster centers are defined for unclassified observations. A class is determined for each cluster. A distance value is computed between a classified observation and each cluster center. When the class of the classified observation is not the class determined for the cluster center having a minimum distance, a first distance value is selected as the minimum distance, a second distance value is selected as the distance value computed to the cluster center having the class of the classified observation, a ratio value is computed between the second distance value and the first distance value, and the class of the classified observation is changed to the class determined for the cluster center having the minimum distance value when the computed ratio value satisfies a label correction threshold. A classification matrix is defined using corrected observations to determine the class for the unclassified observations.

A computing system trains a clustering model. A responsibility parameter vector is initialized for each observation vector and includes a probability value of a cluster membership. The observation vectors include a plurality of classified observation vectors and a plurality of unclassified observation vectors. (A) Beta distribution parameter values are computed for each cluster. (B) Parameter values are computed for a normal-Wishart distribution for each cluster. (C) Each responsibility parameter vector is updated using the beta distribution parameter values, the parameter values, and a respective observation vector. (D) A convergence parameter value is computed. (E) (A) to (D) are repeated until the computed convergence parameter value indicates the responsibility parameter vector defined for each observation vector of the plurality of unclassified observation vectors is converged. A cluster membership is determined and output for each observation vector using a respective, updated responsibility parameter vector.

Distributable clustering model training system

Data is classified using automatically selected hyperparameter values. (A) A first loss value is determined based on a converged classification matrix. (B) Each observation vector is assigned to a cluster using a clustering algorithm based on the converged classification matrix. (C) A predefined number of observation vectors is selected from each cluster. D) Classified observation vectors and unclassified observation vectors are updated based on the selections in (C) and (A) is repeated. (E) An entropy loss value is determined, wherein (A) to (E) are repeated for a plurality of different values of a kernel parameter value and a batch size value. (F) A second loss value is determined based on the converged classification matrix, a label matrix defined from the converged classification matrix, and a weight value. (L) (A) to (F) are repeated with a plurality of different values of the weight value until convergence is satisfied.

Distributed hyperparameter tuning system for active machine learning

A computer trains a classification model. (A) An estimation vector is computed for each observation vector using a weight value, a mean vector, and a covariance matrix. The estimation vector includes a probability value for each class of a plurality of classes for each observation vector that indicates a likelihood that each observation vector is associated with each class. A subset of the plurality of observation vectors has a predefined class assignment. (B) The weight value is updated using the computed estimation vector. (C) The mean vector for each class is updated using the computed estimation vector. (D) The covariance matrix for each class is updated using the computed estimation vector. (E) A convergence parameter value is computed. (F) A classification model is trained by repeating (A) to (E) until the computed convergence parameter value indicates the mean vector for each class of the plurality of classes is converged.

Distributable classification system

A computing device predicts an event or classifies an observation. A trained labeling model is executed with unlabeled observations to define a label distribution probability matrix used to select a label for each observation. Unique combinations of observations selected from the unlabeled observations are defined. A marginal distribution value is computed from the label distribution probability matrix. A joint distribution value is computed between observations included in each combination. A mutual information value is computed for each combination as a combination of the marginal distribution value and the joint distribution value computed for the respective combination. A predefined number of observation vector combinations is selected from the combinations that have highest values for the computed mutual information value. Labeled observation vectors are updated to include each observation vector included in the selected observation vector combinations with a respective obtained label.

Machine learning predictive labeling system

A computing device predicts an event or classifies an observation. A trained labeling model is executed with unlabeled observations to define a label distribution probability matrix. A label is selected for each observation. A mean observation vector and a covariance matrix are computed from the unlabeled observations selected to have each respective label. A number of eigenvalues that have a smallest value is selected from each covariance matrix and used to define a null space for each respective label. A distance value is computed for a distance vector computed to the mean observation vector and projected into the null space associated with the label selected for each respective observation. A diversity rank is determined for each respective observation based on minimum computed distance values. A predefined number of observations having highest values for the diversity rank are included in labeled observations and removed from the unlabeled observations.

A computing device automatically classifies an observation vector. A label set defines permissible values for a target variable. Supervised data includes a labeled subset that has one of the permissible values. A converged classification matrix is computed based on the supervised data and an unlabeled subset using a prior class distribution matrix that includes a row for each observation vector. Each column is associated with a single permissible value of the label set. A cell value in each column is a likelihood that each associated permissible value of the label set occurs based on prior class distribution information. The value of the target variable is selected using the converged classification matrix. A weighted classification label distribution matrix is computed from the converged classification matrix. The value of the target variable for each observation vector of the plurality of observation vectors is output to a labeled dataset.

A computing device automatically classifies an observation vector. (a) A converged classification matrix is computed that defines a label probability for each observation vector. (b) The value of the target variable associated with a maximum label probability value is selected for each observation vector. Each observation vector is assigned to a cluster. A distance value is computed between observation vectors assigned to the same cluster. An average distance value is computed for each observation vector. A predefined number of observation vectors are selected that have minimum values for the average distance value. The supervised data is updated to include the selected observation vectors with the value of the target variable selected in (b). The selected observation vectors are removed from the unlabeled subset. (a) and (b) are repeated. The value of the target variable for each observation vector is output to a labeled dataset.

Growing community of inventors

Apex, NC, United States of America

Xu Chen