-
公开(公告)号：US20140341022A1
公开(公告)日：2014-11-20
申请号：US14451691
申请日：2014-08-05
Applicant: Huawei Technologies Co., Ltd.
Inventor： Lili Wei , Wen Chen , Yuejun Wei , Ying Jin
IPC: H04L12/811 , H04L12/825
CPC classification number: H04L47/38 , H04B7/15521 , H04L1/0002 , H04L47/25 , H04L2001/0097
Abstract: Embodiments of the present invention relate to the field of communications, and provide a network coding method, a relay apparatus and a selection apparatus, which can avoid a case that a network coding system matrix is not full rank, and improve correctness of decoding. The network coding method includes: obtaining network coding information, where the network coding information includes information of a candidate network coding vector set and a candidate transmission rate set, and transmission rates in the candidate transmission rate set are in one-to-one correspondence with network coding vectors in the candidate network coding vector set; selecting a full rank network matrix according to the network coding information; and coding received source node information according to the full rank network matrix. The network coding method, relay apparatus and selection apparatus provided in the embodiments of the present invention are used for network coding.
Abstract translation: 本发明的实施例涉及通信领域，并且提供一种网络编码方法，中继装置和选择装置，其可以避免网络编码系统矩阵不是满秩的情况，并且提高解码的正确性。 网络编码方法包括：获取网络编码信息，其中网络编码信息包括候选网络编码矢量集和候选传输速率集合的信息，并且候选传输速率集合中的传输速率与一 候选网络编码向量集中的网络编码向量; 根据网络编码信息选择完整秩网络矩阵; 并根据满秩网络矩阵对接收的源节点信息进行编码。 在本发明实施例中提供的网络编码方法，中继装置和选择装置用于网络编码。
-
公开(公告)号：US11966844B2
公开(公告)日：2024-04-23
申请号：US17981120
申请日：2022-11-04
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
IPC: G06N3/08 , G06F7/14 , G06F8/71 , G06F18/214 , G06F18/22 , G06N3/044 , G06N3/045 , G06N3/082 , G06N5/04 , G06Q40/00 , G06V20/58
CPC classification number: G06N3/08 , G06F7/14 , G06F8/71 , G06F18/214 , G06F18/22 , G06N3/044 , G06N3/045 , G06N3/082 , G06N5/04 , G06Q40/00 , G06V20/58
Abstract: This application provides a method for training a neural network model and an apparatus. The method includes: obtaining annotation data that is of a service and that is generated by a terminal device in a specified period; training a second neural network model by using the annotation data that is of the service and that is generated in the specified period, to obtain a trained second neural network model; and updating a first neural network model based on the trained second neural network model. In the method, training is performed based on the annotation data generated by the terminal device, so that in an updated first neural network model compared with a universal model, an inference result has a higher confidence level, and a personalized requirement of a user can be better met.
-
Patent Agency Ranking
公开(公告)号：US11521012B2
公开(公告)日：2022-12-06
申请号：US16910289
申请日：2020-06-24
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Abstract: This application provides a method for training a neural network model and an apparatus. The method includes: obtaining annotation data that is of a service and that is generated by a terminal device in a specified period; training a second neural network model by using the annotation data that is of the service and that is generated in the specified period, to obtain a trained second neural network model; and updating a first neural network model based on the trained second neural network model. In the method, training is performed based on the annotation data generated by the terminal device, so that in an updated first neural network model compared with a universal model, an inference result has a higher confidence level, and a personalized requirement of a user can be better met.
-
公开(公告)号：US09521084B2
公开(公告)日：2016-12-13
申请号：US14451691
申请日：2014-08-05
Applicant: Huawei Technologies Co., Ltd.
Inventor： Lili Wei , Wen Chen , Yuejun Wei , Ying Jin
IPC: G01R31/08 , H04L12/811 , H04B7/155 , H04L12/825 , H04L1/00
CPC classification number: H04L47/38 , H04B7/15521 , H04L1/0002 , H04L47/25 , H04L2001/0097
Abstract: Embodiments of the present invention relate to the field of communications, and provide a network coding method, a relay apparatus and a selection apparatus, which can avoid a case that a network coding system matrix is not full rank, and improve correctness of decoding. The network coding method includes: obtaining network coding information, where the network coding information includes information of a candidate network coding vector set and a candidate transmission rate set, and transmission rates in the candidate transmission rate set are in one-to-one correspondence with network coding vectors in the candidate network coding vector set; selecting a full rank network matrix according to the network coding information; and coding received source node information according to the full rank network matrix. The network coding method, relay apparatus and selection apparatus provided in the embodiments of the present invention are used for network coding.
Abstract translation: 本发明的实施例涉及通信领域，并且提供一种网络编码方法，中继装置和选择装置，其可以避免网络编码系统矩阵不是满秩的情况，并且提高解码的正确性。 网络编码方法包括：获取网络编码信息，其中网络编码信息包括候选网络编码矢量集和候选传输速率集合的信息，并且候选传输速率集合中的传输速率与一 候选网络编码向量集中的网络编码向量; 根据网络编码信息选择完整秩网络矩阵; 并根据满秩网络矩阵对接收的源节点信息进行编码。 在本发明实施例中提供的网络编码方法，中继装置和选择装置用于网络编码。
-
公开(公告)号：US20140101498A1
公开(公告)日：2014-04-10
申请号：US14104578
申请日：2013-12-12
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Inventor： Jie XIONG , Yuejun Wei , Ying Jin , Qian Zhu
IPC: G06F11/10
CPC classification number: G06F11/10 , H04L1/0054 , H04L1/203
Abstract: A method for estimating a block error rate and a communication device are applied to the field of communications technologies. The method for estimating a block error rate includes: decoding N received coded code blocks to obtain multiple posterior probabilities APPs, where N is a natural number greater than 1; obtaining, according to the multiple posterior probabilities APPs and a preset policy, a result indicating that the decoding of each coded code block is correct or incorrect, where the preset policy includes: when a sum of absolute values of the multiple APPs is greater than or equal to a preset threshold, the decoding is correct; and obtaining a decoding block error rate according to a result indicating whether the decoding of the N coded code blocks is correct. In this way, the estimation of a decoding block error rate is implemented.
Abstract translation: 用于估计块错误率的方法和通信设备被应用于通信技术领域。 用于估计块错误率的方法包括：解码N个接收到的编码码块以获得多个后验概率APP，其中N是大于1的自然数; 根据多个后验概率APP和预设策略，获得指示每个编码代码块的解码正确或不正确的结果，其中预设策略包括：当多个APP的绝对值之和大于或等于 等于预设阈值，解码正确; 以及根据指示所述N个编码代码块的解码是否正确的结果来获得解码块错误率。 以这种方式，实现了解码块错误率的估计。
-
公开(公告)号：US20200320344A1
公开(公告)日：2020-10-08
申请号：US16910289
申请日：2020-06-24
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Abstract: This application provides a method for training a neural network model and an apparatus. The method includes: obtaining annotation data that is of a service and that is generated by a terminal device in a specified period; training a second neural network model by using the annotation data that is of the service and that is generated in the specified period, to obtain a trained second neural network model; and updating a first neural network model based on the trained second neural network model. In the method, training is performed based on the annotation data generated by the terminal device, so that in an updated first neural network model compared with a universal model, an inference result has a higher confidence level, and a personalized requirement of a user can be better met.
-
公开(公告)号：US09378085B2
公开(公告)日：2016-06-28
申请号：US14104578
申请日：2013-12-12
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Inventor： Jie Xiong , Yuejun Wei , Ying Jin , Qian Zhu
CPC classification number: G06F11/10 , H04L1/0054 , H04L1/203
Abstract: A method for estimating a block error rate and a communication device are applied to the field of communications technologies. The method for estimating a block error rate includes: decoding N received coded code blocks to obtain multiple posterior probabilities APPs, where N is a natural number greater than 1; obtaining, according to the multiple posterior probabilities APPs and a preset policy, a result indicating that the decoding of each coded code block is correct or incorrect, where the preset policy includes: when a sum of absolute values of the multiple APPs is greater than or equal to a preset threshold, the decoding is correct; and obtaining a decoding block error rate according to a result indicating whether the decoding of the N coded code blocks is correct. In this way, the estimation of a decoding block error rate is implemented.
Abstract translation: 用于估计块错误率的方法和通信设备被应用于通信技术领域。 用于估计块错误率的方法包括：解码N个接收到的编码码块以获得多个后验概率APP，其中N是大于1的自然数; 根据多个后验概率APP和预设策略，获得指示每个编码代码块的解码正确或不正确的结果，其中预设策略包括：当多个APP的绝对值之和大于或等于 等于预设阈值，解码正确; 以及根据指示所述N个编码代码块的解码是否正确的结果来获得解码块错误率。 以这种方式，实现了解码块错误率的估计。
-
-
-
-
-
-
Search Results
7
records
(took 0.016 seconds)
