公开(公告)号：US20220164604A1

公开(公告)日：2022-05-26

申请号：US17602282

申请日：2020-03-26

Applicant: NIPPON TELEGRAPH AND TELEPHONE CORPORATION

Inventor： Sekitoshi KANAI ,  Hiroshi TAKAHASHI

IPC: G06K9/62 ,  G06F16/906 ,  G06N3/04 ,  G06N3/08

Abstract: A classification device (10) includes: a classification unit (12) that performs classification by using a model (121) that is a model performing classification and is a deep learning model; and a preprocessing unit (11) that is provided prior to the classification unit (12), and selects an input to the model (121) by using a mask model (111) that minimizes a sum of a loss function and a magnitude of the input to the classification unit (12), the loss function evaluating a relationship between a label on an input from teaching data and an output of the model (121).

公开(公告)号：US20220405624A1

公开(公告)日：2022-12-22

申请号：US17777074

申请日：2019-11-21

Applicant: NIPPON TELEGRAPH AND TELEPHONE CORPORATION

Inventor： Hiroshi TAKAHASHI ,  Tomoharu IWATA ,  Sekitoshi KANAI ,  Atsutoshi KUMAGAI ,  Yuki YAMANAKA ,  Masanori YAMADA ,  Satoshi YAGI

IPC: G06N7/00 ,  G06N20/00

Abstract: An acquisition unit 15a acquires data in a task. The learning unit 15b learns a generation model representing a distribution of a probability of the data in the task so that a mutual information amount between a latent variable and an observed variable is minimized in the model.

公开(公告)号：US20250155277A1

公开(公告)日：2025-05-15

申请号：US18725072

申请日：2022-01-11

Applicant: NIPPON TELEGRAPH AND TELEPHONE CORPORATION

Inventor： Yoshifumi WAKISAKA ,  Hiroshi TAKAHASHI ,  Daisuke IIDA ,  Yusuke KOSHIKIYA

IPC: G01H9/00

Abstract: In order to achieve the above object, a signal processing method according to the present disclosure includes: a calculation step of calculating a phase change at each of points on a medium from a phase of a reflected signal generated by a pulse signal transmitted to the medium, at each of predetermined times; a phase connection step of performing phase connection processing, at each of the predetermined times, on the phase change at each of the points calculated in the calculation step; and a correction step of detecting an outlier using a phase change at each of the points at the same time at which the phase connection processing is performed in the phase connection step, and correcting the outlier, at each of the predetermined times.

公开(公告)号：US20210264285A1

公开(公告)日：2021-08-26

申请号：US17253131

申请日：2019-06-19

Applicant: NIPPON TELEGRAPH AND TELEPHONE CORPORATION

Inventor： Hiroshi TAKAHASHI ,  Tomoharu IWATA ,  Yuki YAMANAKA ,  Masanori YAMADA ,  Satoshi YAGI

IPC: G06N3/08 ,  G08B5/22 ,  H04Q9/00 ,  G06N3/04

Abstract: An acquisition unit (15a) acquires data output by sensors. A learning unit (15b) substitutes a prior distribution of an encoder in a generative model including the encoder and a decoder and representing a probability distribution of the data with a marginalized posterior distribution that marginalizes the encoder, approximates a Kullback-Leibler information quantity using a density ratio between a standard Gaussian distribution and the marginalized posterior distribution, and learns the generative model using data. A detection unit (15c) estimates a probability distribution of the data using the learned generative model and detects an event in that an estimated occurrence probability of the data newly acquired is lower than a prescribed threshold as abnormality.