-
公开(公告)号:WO2021191556A1
公开(公告)日:2021-09-30
申请号:PCT/FR2021/050495
申请日:2021-03-23
Applicant: SAFRAN
Inventor: CHAPDELAINE, Camille Jocelyn Roger , PICARD, Sylvaine
IPC: G06N3/04 , G06N3/08 , G06K9/00 , G06T7/00 , G06K2209/19 , G06K9/6274 , G06N20/10 , G06N3/0454
Abstract: Un aspect de l'invention concerne un procédé d'apprentissage d'une pluralité des réseaux de neurones se repoussant entre eux destiné au contrôle de pièces mécaniques à l'aide d'une méthode itérative permettant de rendre la mise à jour des neurones linéairement dépendante du nombre de réseaux de neurones tout en garantissant l'obtentions de réseaux de neurones se repoussant entre eux bien répartis sur la distribution de leurs paramètres connaissant les données de travail. Un deuxième aspect de l'invention concerne un procédé de contrôle de pièce mécanique faisant usage d'un procédé selon un premier aspect de l'invention.
-
公开(公告)号:WO2022012144A1
公开(公告)日:2022-01-20
申请号:PCT/CN2021/094023
申请日:2021-05-17
Applicant: 湖南大学
IPC: G06K9/62 , G06K9/6223 , G06K9/6274 , G06N3/006 , G06N3/0454 , G06N3/08
Abstract: 本发明公开了一种基于不平衡数据深度信念网络的并行入侵检测方法,其读取不平衡数据集数据,对不平衡数据采用改进的NCL算法进行欠采样处理,降低多数类样本的比重,使数据集数据分布均衡;在分布式内存计算平台Spark平台上采用改进的差分进化算法对深度信念网络模型的参数进行优化,得到最优的模型参数;对数据集数据进行特征提取,然后采用加权后的核极限学习机进行入侵检测分类,最后通过多线程并行的训练多个不同结构的加权后的核极限学习机作为基分类器,建立基于自适应加权投票的多分类器入侵检测模型进行并行入侵检测。本发明能解决现有入侵检测方法对不平衡数据集缺乏针对性、训练时间过长的技术问题,并提高优化深度信念网络模型参数的速度。
-
公开(公告)号:WO2021252145A1
公开(公告)日:2021-12-16
申请号:PCT/US2021/032705
申请日:2021-05-17
Applicant: QUALCOMM INCORPORATED
Inventor: LIU, Peng , WANG, Lei , WANG, Zhen , CHENG, Ke-Li , BI, Ning
IPC: G06K9/00 , G06K9/46 , G06K9/62 , G06K9/6256 , G06K9/6274 , G06T11/001 , G06T17/20 , G06T19/20 , G06T2207/20084 , G06T2207/20132 , G06T2207/30201 , G06T2219/2016 , G06T7/50 , G06T7/70 , G06V10/454 , G06V20/64 , G06V40/161 , G06V40/171
Abstract: Systems and techniques are provided for facial image augmentation. An example method can include obtaining a first image capturing a face. Using the first image, the method can determine, using a prediction model, a UV face position map including a two-dimensional (2D) representation of a three-dimensional (3D) structure of the face. The method can generate, based on the UV face position map, a 3D model of the face. The method can generate an extended 3D model of the face by extending the 3D model to include region(s) beyond a boundary of the 3D model. The region(s) can include a forehead region, a region surrounding at least a portion of the face, and/or other region. The method can generate, based on the extended 3D model, a second image depicting the face in a rotated position relative to a position of the face in the first image.
-
公开(公告)号:WO2021259870A1
公开(公告)日:2021-12-30
申请号:PCT/EP2021/066867
申请日:2021-06-21
Applicant: CARTESIAM
Inventor: DE ROCHEBOUET, François , MOUSSA, Mohamed-Ali
IPC: G06F17/18 , G06N20/00 , G06K9/6215 , G06K9/6259 , G06K9/6274 , G06V2201/06
Abstract: Un aspect de l'invention concerne un procédé d'apprentissage pour la détection d'anomalies sur microcontrôleur comportant une mémoire stockant un nombre prédéfini de catégories et recevant des jeux de données multivariables d'un capteur, comportant les étapes suivantes : Calcul d'une moyenne et d'une matrice de covariance pour un ensemble de jeux de données; Tant que la matrice de covariance est mal conditionnée : Ajout d'un jeu de données dans l'ensemble et mise à jour de la moyenne et de la matrice de covariance; Création d'une catégorie associée à la moyenne et à la matrice de covariance dans la mémoire; Pour chaque catégorie, calcul d'une mesure de distance entre la catégorie et chaque autre catégorie; Sélection des catégories correspondant à la première mesure de distance minimale et fusion des deux catégories sélectionnées.
-
公开(公告)号:WO2021242584A1
公开(公告)日:2021-12-02
申请号:PCT/US2021/033106
申请日:2021-05-19
Applicant: PAYPAL, INC.
Inventor: ZHANG, Jiyi
IPC: G06K9/00 , G06K9/62 , A61B5/00 , G06N3/08 , H01J49/00 , G06K9/6256 , G06K9/6274 , G06N3/0454 , G06T1/005 , G06T2201/0063 , G06V10/454 , G06V10/82
Abstract: Systems, methods, and computer program products for determining an attack on a neural network. A data sample is received at a first classifier neural network and at a watermark classifier neural network, wherein the first classifier neural network is trained using a first dataset and a watermark dataset. The first classifier neural network determines a classification label for the data sample. A watermark classifier neural network determines a watermark classification label for the data sample. A data sample is determined as an adversarial data sample based on the classification label for the data sample and the watermark classification label for the data sample.
-
公开(公告)号:WO2021259806A1
公开(公告)日:2021-12-30
申请号:PCT/EP2021/066675
申请日:2021-06-18
Applicant: SIGNIFY HOLDING B.V.
Inventor: SAVERA, Arhum , MAHDIZADEHAGHDAM, Shahin , MURTHY, Abhishek
IPC: G06K9/00 , G06K9/32 , G06K9/62 , G06K9/6274 , G06K9/6288 , G06V10/143 , G06V10/25 , G06V10/454 , G06V20/52
Abstract: A method for optimizing a neural network is provided, including: (1) capturing, via a first sensor group having a first field of view, a first sample set having a first sensor domain corresponding to the first field of view; (2) capturing, via a second sensor group having a second field of view, a second sample set having a second sensor domain corresponding to the second field of view; (3) generating regions of interest of the second sample set; (4) translating the regions of interest to the first sensor domain; (5) identifying nodes of the neural network which correspond to the translated regions; and (6) optimizing the neural network by at least one of (a) increasing the weight value of the nodes corresponding to the one or more translated regions and (b) decreasing the weight value of the nodes not corresponding to the one or more translated regions.
-
公开(公告)号:WO2021250564A1
公开(公告)日:2021-12-16
申请号:PCT/IB2021/055029
申请日:2021-06-08
Applicant: PICA GROUP S.P.A.
Inventor: LAZZARA, Daniele , MINOTTI, Enrico , MARTINI, Manolo
IPC: G06F21/32 , G06F21/62 , H04L29/06 , H04L9/32 , G06F16/58 , G06K9/00 , G06F21/6245 , G06K9/6274 , G06V40/168 , G06V40/172 , G06V40/40 , H04L63/0861
Abstract: The invention relates in general to a method (1000) for accessing multimedia content comprising the steps of: generating (S1100) a first code (3100), acquiring (S1200) multimedia content (3200), acquiring (S1300) the first code (3100), acquiring (S1400) a recognition image, extracting (S1500) identification data of the recognition image, searching (S1600, S5600) for multimedia content associated with the identification data, and enabling (S1700) access to the multimedia content resulting from the step of searching (S1600).
-
公开(公告)号:WO2021190137A1
公开(公告)日:2021-09-30
申请号:PCT/CN2021/074501
申请日:2021-01-29
Applicant: ALIPAY LABS (SINGAPORE) PTE. LTD.
Inventor: ZHANG, Xingwen , YANG, Shuang
IPC: G01C21/34 , G06N3/08 , G06N20/00 , G01C21/343 , G01C21/3446 , G06K9/00791 , G06K9/6256 , G06K9/6274 , G06N3/006 , G06N3/0445 , G06N3/0454 , G06N5/003
Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining routing. An exemplary method comprises: inputting a plurality of to-be-optimized routing solution candidates to a Siamese neural network comprising a plurality of value prediction networks, each of the value prediction networks being trained to predict a cost associated with a to-be-optimized routing solution candidate; identifying one or more to-be-optimized routing solution candidates from the plurality of to-be-optimized routing solution candidates based on outputs of the Siamese neural network; inputting the one or more identified to-be-optimized routing solution candidates to a routing optimizer to obtain one or more optimized routing solution candidates; and determining an optimized routing solution with a lowest cost from the one or more optimized routing solution candidates.
-
公开(公告)号:WO2021139249A1
公开(公告)日:2021-07-15
申请号:PCT/CN2020/118524
申请日:2020-09-28
Applicant: 平安科技(深圳)有限公司
IPC: G06F11/30 , G06F11/3065 , G06K9/6256 , G06K9/6274
Abstract: 一种数据异常检测方法、装置、设备及存储介质,涉及大数据领域,该方法包括:获取未标记数据(S1);根据预设的查询策略从所述未标记数据中提取出初级异常数据(S2);将所述初级异常数据进行识别标记后存入已标记的第一数据集合中组成第二数据集合,并通过所述第二数据集合对预先训练的超球体分类模型进行训练(S3);识别所述超球体分类模型是否达到训练终止条件(S4);当达到所述训练终止条件,将所述未标记数据输入训练终止条件下的所述超球体分类模型中进行分类筛选,以得到目标异常数据(S5)。该方法利用少量已标记数据训练分类模型,达到训练终止条件后利用该分类模型对未标记数据进行分类,对数据的原始分布没有限制,减少了运营人员需要标记的数据量,分类结果准确度高。
-
公开(公告)号:WO2021121548A1
公开(公告)日:2021-06-24
申请号:PCT/EP2019/085417
申请日:2019-12-16
Applicant: TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
Inventor: HUNT, Alexander , CALTENCO, Hector , BASTANI, Saeed
IPC: G06T3/40 , G06T5/00 , G06K9/00 , G06K9/00671 , G06K9/00771 , G06K9/3233 , G06K9/4628 , G06K9/6274 , G06T2207/20012
Abstract: An image filtering arrangement (100) comprising a controller (101) configured to receive (212) an image data file; propose (220) zero or more regions of interest (ROI) for the image data file; and to select (230) adaptive filtering for at least one of the proposed zero or more regions of interest (ROI) and apply the selected adaptive filtering to the at least one of the proposed zero or more regions of interest (ROI)
-
-
-
-
-
-
-
-
-