Aspects of the invention include detecting an error alert from a target computer system. In response to detecting the error alert, performance data is then retrieved from the target computer system. A gated recurrent unit (GRU) neural network is used to generate a prediction of a root cause of the error alert based on the performance data. The weights of a reset gate of the GRU neural network are adjusted based on received feedback of the prediction.

Techniques for monitoring a computing environment for anomalous activity are presented. An example method includes receiving a request to invoke an action within the computing environment. An anomaly score is generated for the received request by applying a probabilistic model to properties of the request. The anomaly score generally indicates a likelihood that the properties of the request correspond to historical activity within the computing environment for a user associated with the request. The probabilistic model generally comprises a model having been trained using historical activity within the computing environment for a plurality of users, the historical activity including information identifying an action performed in the computing environment and contextual information about a historical request. Based on the generated anomaly score, one or more actions are taken to process the request such that execution of requests having anomaly scores indicative of unexpected activity may be blocked pending confirmation.

Systems and methods are provided framework for automatically choosing the appropriate generalized linear model (GLM) given a time series of count data, and for anomaly detection on time series data. A dispersion parameter is determined and used to determine whether the count data is overdispersed data or underdispersed data. The overdispersed data or the underdispersed data is used to determine a GLM to apply on the dataset. Using the determined GLM on the data, anomalies can be determined.

A data analysis device 10 comprises: a frequency analysis unit 11 that performs frequency analysis, under a predetermined condition, on each piece of a plurality of training data pieces including a plurality of class training data pieces some of which have been assigned a label indicating the data class; a cluster analysis unit 12 that clusters the frequency analyzed training data pieces into a number of classes of frequency analyzed training data; a computation unit 13 that computes, on the basis of the clusters, the degree to which frequency analyzed training data pieces assigned the same label are not included in the same cluster; and a selection unit 14 that selects, as a clustering model for assigning a label to a training data piece, clustering results according to the cluster analysis unit 12 when the smallest degree was computed, from among the plurality of computed degrees.

A method is described for extracting aperiodic components from a time-series wave data set for diagnosis purposes. The method may include collecting time-series wave data within a controlled environment were a plurality of contrasting conditions can be used in collecting the time-series wave data set. Aperiodic components can be extracted from the time-series wave data set and the aperiodic components can then be fitted to the plurality of contrasting conditions of the controlled environment to product regressed aperiodic components from which diagnostic determination can be made.

The coordinate system fixing unit uses the displacement of an object under measurement between photographed images in chronological order to generate fixed-coordinate chronological images. The displacement calculation unit uses the fixed-coordinate chronological images to calculate a two-dimensional spatial distribution of the displacement of the surface of the object under measurement. The displacement difference calculation unit calculates a two-dimensional displacement difference distribution by removing an error component from the two-dimensional spatial distribution. The depth movement amount calculation unit calculates a depth movement amount from the two-dimensional displacement difference distribution. The displacement separation unit calculates in-plane displacement from the two-dimensional displacement difference distribution. The determination unit uses the in-plane displacement and/or the depth movement amount to determine whether there is an abnormality in the object under measurement.

The present disclosure relates to a method performed by a people distinguishing system (1) for in an image distinguishing human beings in a crowd. The people distinguishing system identifies (1001) one or more detected objects classified as human beings (2) in an image (3) derived from a thermal camera (10) adapted to capture a scene in an essentially forward-looking angle. The people distinguishing system further identifies at least a first grouping (4) of adjoining pixels in the image, not comprised in the one or more detected human beings, having an intensity within a predeterminable intensity range. Moreover, the people distinguishing system determines (1003) a grouping pixel area (40) of the at least first grouping in the image. Furthermore, the people distinguishing system determines (1004) for at least a first vertical position (y.sub.expected) in the image, based on head size reference data (5), an expected pixel area (x.sub.expected) of a human head at the at least first vertical position. The people distinguishing system further compares (1005) at least a portion of the grouping pixel area with the expected head pixel area for the at least first vertical position. Moreover, the people distinguishing system determines (1006) that the at least first grouping comprises at least a first overlapping human being (6), when at least a first comparison resulting from the comparing exceeds a predeterminable conformity threshold. The disclosure also relates to a people distinguishing system in accordance with the foregoing, a thermal camera comprising such people distinguishing system, and a respective corresponding computer program product and non-volatile computer readable storage medium.

A model analyzer may receive a representative data set as input and select one of a plurality of analytic models to perform the analysis. Before deciding which model to use the model may be trained, and the trained model evaluated for accuracy. However, some models are known to behave poorly when the training data is distributed in a particular way. Thus, the cost of training a model and evaluating the trained model can be avoided by first analyzing the distribution of the representative data. Identifying the representative data distribution allows ruling out use of models for which the distribution of the representative data is unsuitable. Only models that may be compatible with the distribution of the representative data may be trained and evaluated for accuracy. The most accurate trained model whose accuracy meets an accuracy threshold may be selected to analyze subsequently received data related to the representative data.

Techniques are provided for extracting anomaly related rules from organizational data. One method comprises obtaining anomaly analysis data integrated from multiple data sources of an organization, wherein the multiple data sources comprise at least one set of labeled anomaly data related to anomalous transactions; extracting features from the integrated anomaly analysis data that correlate with an indication of an anomaly; training multiple machine learning models using the extracted features, where the machine learning models are trained using different combinations of the extracted features; evaluating a performance of the trained machine learning models; and extracting rules from the trained machine learning models based on the performance, wherein the extracted rules are used to classify transactions as anomalous. The trained machine learning models comprise a decision tree comprising paths to an anomaly classification. The extracted rules are optionally in a human-readable format.

The technology disclosed relates to classification of process cycle images to predict success or failure of process cycles. The technology disclosed includes capturing and processing images of sections arranged on an image generating chip in genotyping process. Image description features of production cycle images are created and given as input to classifiers. A trained classifier separates successful production images from unsuccessful or failed production images. The failed production images are further classified by a trained root cause classifier into various categories of failure.