Abstract: An example method for updating convolutional neural network includes: obtaining a sample with a classification label; performing a first operation on the sample based on parameters of each layer of front-end network, to obtain a first operation result; performing a second operation on the sample based on the first operation result and the parameters of each layer of back-end network that the first GPU has, to obtain a second operation result; separately sending the first operation result to the other GPUs; receiving a third operation result obtained after each other GPU performs a third operation on the sample based on their parameters of each layer of back-end network and the first operation result; combining the second and third operation results to obtain a classification result; determining a prediction error based on the classification result and the classification label; and updating the convolutional neural network based on the prediction error.

Abstract: The present specification discloses a trusted execution environment (TEE)-based model training method and apparatus. In one or more embodiments, the method includes: obtaining encrypted target samples from an encrypted training sample set in a first execution environment, inputting the encrypted target samples into a second execution environment that is a trusted execution environment (TEE) different from the first execution environment, decrypting the encrypted target samples in the TEE to obtain decrypted target samples, inputting the decrypted target samples into a feature extraction model in the TEE to determine sample features, determining the sample features output from the TEE as target sample features for a current iteration of a training process for a target model, and performing, based on the target sample features, the current iteration on the target model in the first execution environment.

Abstract: The present specification discloses a trusted execution environment (TEE)-based model training method and apparatus. In one or more embodiments, the method includes: obtaining encrypted target samples from an encrypted training sample set in a first execution environment, inputting the encrypted target samples into a second execution environment that is a trusted execution environment (TEE) different from the first execution environment, decrypting the encrypted target samples in the TEE to obtain decrypted target samples, inputting the decrypted target samples into a feature extraction model in the TEE to determine sample features, determining the sample features output from the TEE as target sample features for a current iteration of a training process for a target model, and performing, based on the target sample features, the current iteration on the target model in the first execution environment.

Abstract: Implementations of the present specification provide a method, an apparatus, and a device for updating a convolutional neural network by using a GPU cluster. The GPU cluster includes a first GPU and several other GPUs.

Abstract: A method to provide privacy protection on a video includes performing people recognition on the video, when a registered person is recognized, determining if a privacy setting is turned on, and, when the privacy setting is turned on, performing privacy protection on the video. Performing privacy protection on the video includes pausing a recording of a live video, obscuring the registered person in the live video or a recorded video, or skipping ahead or deleting part of the recorded video.

Abstract: A method to create a photo diary includes creating an entry for a time period in the photo diary. Creating the entry includes selecting candidate frames from digital videos created or received in the time period. Selecting the candidate frames includes performing people and pet detection on the digital videos, extract frames with people and pets from the digital videos perform people recognition on the frames with people to identify frames with recognized persons. The method further includes sorting the candidate frames with recognized persons and pets into groups based on criteria, selecting representative frames from the groups, and presenting the representative frames in the entry.

Abstract: A method to create a photo diary includes creating an entry for a time period in the photo diary. Creating the entry includes selecting candidate frames from digital videos created or received in the time period. Selecting the candidate frames includes performing people and pet detection on the digital videos, extract frames with people and pets from the digital videos perform people recognition on the frames with people to identify frames with recognized persons. The method further includes sorting the candidate frames with recognized persons and pets into groups based on criteria, selecting representative frames from the groups, and presenting the representative frames in the entry.

Abstract: A method to provide privacy protection on a video includes performing people recognition on the video, when a registered person is recognized, determining if a privacy setting is turned on, and, when the privacy setting is turned on, performing privacy protection on the video. Performing privacy protection on the video includes pausing a recording of a live video, obscuring the registered person in the live video or a recorded video, or skipping ahead or deleting part of the recorded video.

Abstract: A device includes a link interface circuit, a first plurality of allocated buffers, and a second plurality of non-allocated buffers. The link interface circuit is operable to communicate over a communications link using a plurality of virtual channels. A different subset of the plurality of allocated buffers is allocated to each of the virtual channels. The non-allocated buffers are not allocated to a particular virtual channel. The link interface circuit is operable to receive a first transaction over the communications link and assign the first transaction to one of the allocated buffers or one of the non-allocated buffers.

Abstract: A device includes a link interface circuit, a first plurality of allocated buffers, and a second plurality of non-allocated buffers. The link interface circuit is operable to communicate over a communications link using a plurality of virtual channels. A different subset of the plurality of allocated buffers is allocated to each of the virtual channels. The non-allocated buffers are not allocated to a particular virtual channel. The link interface circuit is operable to receive a first transaction over the communications link and assign the first transaction to one of the allocated buffers or one of the non-allocated buffers.