Nec Corporation

Other

Yale University

Jun-ichi Takeuchi

Andrew Ross Barron

Kenji Yamanishi

An outlier detection device for detecting abnormal data in a data set an outlier rule preservation unit, a filtering unit, a degree of outlier calculation unit, a sampling unit, and a supervised learning unit. The supervised learning unit generates a new rule characterizing abnormal data by supervised learning based on a set of the respective data to which a label indicating whether data is abnormal data or not is applied. The new rule is added to the set of rules held in the outlier rule preservation unit to update the rules. The filtering unit uses the rules for determining whether each data of the data set is abnormal or not. The abnormal data is compared to normal data for calculating a degree of abnormality by the degree of outlier calculation unit.

Outlier determination rule generation device and outlier detection device, and outlier determination rule generation method and outlier detection method thereof

Degree of outlier of one input data is calculated by an amount of change in a learned probability density from that before learning as a result of taking in of the input data. This is because data largely differing in a tendency from a so far learned probability density function can be considered to have a high degree of outlier. More specifically, a function of a distance between probability densities before and after data input is calculated as a degree of outlier. Accordingly, a probability density estimation device appropriately estimates a probability distribution of generation of unfair data while sequentially reading a large volume of data and a score calculation device calculates and outputs a degree of outlier of each data based on the estimated probability distribution.

Degree of outlier calculation device, and probability density estimation device and forgetful histogram calculation device for use therein

A time-series model learning unit reads in time-series data sequentially, learns the parameters of a time-series model and stores the parameters in a storage device. A loss function calculating unit reads in sequentially from storage each item of the input time-series data one at a time and calculates values of a loss function. A complexity calculating unit sums the loss values to sequentially calculate complexity as fitting error resulting when a time-series model is fit to the input data. Complexity is stored. A change-point searching unit reads in complexity regarding time-series data before and after change-point candidates from storage with respect to all change-point candidates, compares the sum of the complexities with complexity regarding all time-series data in a case where a change point is assumed to be absent, calculates the change-point score based upon the difference between the compared values, and detects and outputs the change point.

Change-point detection apparatus, method and program therefor

In a Bayes mixture probability density calculator for calculating Bayes mixture probability density which reduces a logarithmic loss A modified Bayes mixture probability density is calculated by mixing traditional Bayes mixture probability density calculated on given model S with a small part of Bayes mixture probability density for exponential fiber bundle on the S. Likewise, a prediction probability density calculator is configured by including the Bayes mixture probability density calculator, and by using Jeffreys prior distribution in traditional Bayes procedure on the S.

Device, method, and medium for predicting a probability of an occurrence of a data

In a Bayes mixture probability density calculator for calculating Bayes mixture probability density which reduces a logarithmic loss A modified Bayes mixture probability density is calculated by mixing raditional Bayes mixture probability density calculated on given model S with a small pary of Bayes mixture probability density for exponential fiber bundle on the S. Likewise, a prediction probability calculator is configured by including the Bayes mixture probability density calculator, and by using Jeffredys prior distribution in traditional Bayes procedure on the S.

Growing community of inventors

Tokyo, Japan

Jun-ichi Takeuchi