A non-transitory computer-readable recording medium storing a program that causes a computer to execute a process, the process includes generating, based on an index value related to an evaluation function value, a first candidate target from combinatorial targets in a combinatorial optimization problem that minimizes the evaluation function value under a plurality of constraint conditions, analyzing, based on a first result obtained by solving and optimizing based on the first candidate target, a combination that is included in the first result and that is a constraint violation, selecting, from among the combinatorial targets, a target related to resolving of the constraint violation that has been analyzed, obtaining, based on a second candidate target that include the selected combinatorial target and the first result, a second result which is optimized, and determining a solving result of the combinatorial optimization problem based on an evaluation result of the second result.

Various embodiments of a computer-implemented system which learns textual representations while filtering out potentially personally identifying data and retaining semantic meaning within the textual representations are disclosed herein.

A data-driven PMU bad data detection algorithm based on spectral clustering using single PMU data is described. The described algorithm does not require the system topology and parameters. First, a data identification method based on a decision tree is described to distinguish event data and bad data by using the slope feature of each set of data. Then, a bad data detection method based on spectral clustering is described. By analyzing the weighted relationships among all the data, this method can detect the bad data that has a small deviation.

A data-driven PMU bad data detection algorithm based on spectral clustering using single PMU data is described. The described algorithm does not require the system topology and parameters. First, a data identification method based on a decision tree is described to distinguish event data and bad data by using the slope feature of each set of data. Then, a bad data detection method based on spectral clustering is described. By analyzing the weighted relationships among all the data, this method can detect the bad data that has a small deviation.

Various embodiments of a computer-implemented system which learns textual representations while filtering out potentially personally identifying data and retaining semantic meaning within the textual representations are disclosed herein.

The present disclosure relates generally to machine learning. More particularly, the present disclosure relates to systems and methods that regularize neural networks by decorrelating neurons or other parameters of the neural networks during training of the neural networks promoting these parameter to innovate over one another.

A method of hop count matrix recovery based on a decision tree classifier, includes: S1: performing a flooding process to acquire a hop count matrix {tilde over (H)} with missing entries; S2: constructing a training sample set according to relationships between a part of observed hop counts in the hop count matrix {tilde over (H)}, and modeling the observed hop counts in the hop count matrix as labels of the training sample set, wherein a maximum hop count represents a number of classes; S3: training a decision tree classifier according to the training sample set obtained in step S2; and S4: constructing a feature for an unobserved hop count, to obtain an unknown sample; and inputting the unknown sample to the trained decision tree classifier, to obtain a class of the unknown sample which represents a missing hop count at a corresponding position in the matrix, to recover a complete hop count matrix H.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a convolutional neural network using a regularization scheme. One of the methods includes repeatedly performing the following operations: obtaining a kernel of a particular convolutional layer; applying a Fourier transform to the kernel; generating a decomposition using singular-value decomposition (SVD); generating a regularized diagonal matrix; generating a recomposition; applying an inverse Fourier transform to the recomposition; and training the convolutional neural network on training inputs.

According to one embodiment, a data analyzing apparatus acquires data containing the number N of analysis target samples (where N is an integer larger than or equal to 2). The apparatus performs a matrix factorization upon the data to factorize the data into the number K of basis samples and the number K of weights corresponding to the number K of basis samples (where K is an integer larger than or equal to 2), and fixes part of the K basis samples to specific basis samples in the matrix factorization.

A method of building a model for estimating a level of psychological safety includes acquiring, by a computer, post data communicated between members in a team, identifying fixed type post data that does not contribute to evaluation of the psychological safety among the acquired post data, and creating the model based on content of the acquired post data from which the fixed type post data has been removed.