Abstract: According to one embodiment, a behavior estimating method for estimating behavior of a person being measured equipped with an activity meter including an acceleration sensor, includes presetting a first angle between a gravitation direction and a direction of action of a standard person being measured, calculating different acceleration waveforms from the activity meter, extracting a second angle between the gravitation direction and the direction of action of the person from the acceleration waveforms, performing a coordinate transformation process among the acceleration waveforms based upon the present first angle and the extracted second angle, and performing behavior estimation of the person using a result of the coordinate transformation.

Abstract: According to one embodiment, an operation estimating method comprises presetting a plurality of operation candidates to be estimated, acquiring acceleration data from an activity meter worn by a person being measured, calculating estimated probabilities of the operation candidates from the acquired acceleration data, selecting an operation candidate corresponding to a maximum probability from the calculated estimated probabilities, as an estimated operation, and correcting the estimated operation in consideration of a past operation performance when the maximum probability of the estimated probabilities is lower than a predetermined value.

Abstract: An information display method includes converting P (P?2) high-dimensional data respectively obtained from P images each tagged with a label representing a result of classification into one of a plurality of classes to respectively acquire P low-dimensional data, generating a plot diagram including P plot points obtained by visualizing the P low-dimensional data each depending on the classification result, generating auxiliary information corresponding to a base point set as one of the plot points included in the plot diagram and N (1?N<P) reference points respectively set as the plot points corresponding to the N low-dimensional data extracted from among the low-dimensional data, belonging to a same class as a class to which the base point belongs, among the P low-dimensional data, and generating a display image including the plot diagram to which the auxiliary information is added.

Abstract: An information display method includes converting P (P?2) high-dimensional data respectively obtained from P images each tagged with a label representing a result of classification into one of a plurality of classes to respectively acquire P low-dimensional data, generating a plot diagram including P plot points obtained by visualizing the P low-dimensional data each depending on the classification result, generating auxiliary information corresponding to a base point set as one of the plot points included in the plot diagram and N (1?N<P) reference points respectively set as the plot points corresponding to the N low-dimensional data extracted from among the low-dimensional data, belonging to a same class as a class to which the base point belongs, among the P low-dimensional data, and generating a display image including the plot diagram to which the auxiliary information is added.

Abstract: According to one embodiment, an operation estimating method comprises presetting a plurality of operation candidates to be estimated, acquiring acceleration data from an activity meter worn by a person being measured, calculating estimated probabilities of the operation candidates from the acquired acceleration data, selecting an operation candidate corresponding to a maximum probability from the calculated estimated probabilities, as an estimated operation, and correcting the estimated operation in consideration of a past operation performance when the maximum probability of the estimated probabilities is lower than a predetermined value.

Abstract: According to one embodiment, a behavior estimating method for estimating behavior of a person being measured equipped with an activity meter including an acceleration sensor, includes presetting a first angle between a gravitation direction and a direction of action of a standard person being measured, calculating different acceleration waveforms from the activity meter, extracting a second angle between the gravitation direction and the direction of action of the person from the acceleration waveforms, performing a coordinate transformation process among the acceleration waveforms based upon the present first angle and the extracted second angle, and performing behavior estimation of the person using a result of the coordinate transformation.