-
1.发明授权公开(公告)号：US11995155B2
公开(公告)日：2024-05-28
申请号：US17542692
申请日：2021-12-06
Applicant: Shenzhen Institutes of Advanced Technology
IPC: G06F18/00 , G06F18/2413 , G06F21/36 , G06N3/045
CPC classification number: G06F18/2413 , G06F21/36 , G06N3/045
Abstract: An adversarial image generation method, a computer device, and a computer-readable storage medium are provided. The method includes the following. A reference model classification-equivalent with a target classification model is generated according to the target classification model. A target image is obtained and an original noise for the target image is generated according to the reference model. A first noise and the original noise are input into an adversarial model and a second noise corresponding to the first noise is output when the adversarial model meets a convergence condition, where the second noise enhances an information entropy of the original noise. An enhanced noise image corresponding to the target image is generated according to the second noise and the target image, where a classification accuracy of the enhanced noise image in the target classification model is less than a classification accuracy of the target image in the target classification model.
-
公开(公告)号：US11797372B2
公开(公告)日：2023-10-24
申请号：US17618758
申请日：2020-03-26
Applicant: Shenzhen Institutes of Advanced Technology
Inventor： Lei Peng , Junnan Zhang , Huiyun Li
CPC classification number: G06F11/0793 , G06F11/0721
Abstract: A method and an apparatus for generating time series data based on multi-condition constraints, and a medium are provided. The method include the following: a data repair request from a client is received, the data repair request including to-be-repaired data and condition information; normalization processing is performed on the to-be-repaired data to obtain normalized data, and tensor processing is performed on the condition information to obtain a feature label; a trained data repair model is called to perform repair processing on the normalized data according to the feature label to obtain first repaired data, the data repair model being obtained by training according to sample data, a first sample condition, real sample data and a second sample condition, and the sample data being noise data; and the first repaired data is sent to the client. By adopting embodiments of the application, rich features of the data can be acquired without a large amount of historical data or sample data with high matching degree as training basis, thus ensuring the accuracy and timing of the repaired data, and improving the repair efficiency and quality.
-
公开(公告)号：US11295619B2
公开(公告)日：2022-04-05
申请号：US17121141
申请日：2020-12-14
Applicant: Shenzhen Institutes of Advanced Technology
Inventor： Lei Peng , Yuqiang Sun , Huiyun Li
Abstract: A parking lot data repair method and apparatus, a device, and a storage medium. The method comprises: calculating spatial similarity between two parking lots according to the spatial features of the parking lots and a spatial similarity measurement model; clustering the parking lots into different parking lot clusters according to the spatial similarity, and calculating a spatial similarity probability corresponding to each parking lot cluster; calculating data similarity between two sample parking lots in the same parking lot cluster, and according to the data similarity and the spatial similarity probability corresponding to the parking lot cluster, calculating the data similarity probability of the parking lot cluster under a similar spatial condition; and when the data similarity probability under the similar spatial condition exceeds a probability threshold value, performing data repair on the parking lot to be repaired in the parking lot cluster by means of a cyclic generation-type confrontation network.
-
-
Search Results
3
records
(took 0.024 seconds)
