Boe Technology Group Co., Ltd.

Beijing Boe Display Technology Co., Ltd.

International Business Machines Corporation

Beijing Boe Optoelectronics Technology Co., Ltd.

Tencent Technology (shenzhen) Company Limited

Beijing Boe Technology Development Co., Ltd.

Microsoft Technology Licensing, LLC

Ordos Yuansheng Optoelectronics Co., Ltd.

Chinese Academy of Medical Sciences

Shenzhen Mindray Bio-medical Electronics Co., Ltd.

Beijing Mindray Medical Instrument Co., Ltd.

Xiao Lin Sun

Yun Qiu

Sha Liu

Zhaokun Yang

Xiang Feng

Qiang Zhang

Weipin Hu

Jiuxia Yang

Jin Ping Han

Yiming Zhao

Bing Bai

Feng Bai

Ruizhi Yang

Congcong Wei

Mingxiao Jiang

Paul Michael Solomon

Martin Michael Frank

Dan Wang

Chun Long Wang

Qianqian Bu

Takashi Ando

Vijay Narayanan

Jin-Ping Han

Yulong Li

Dennis Merton Newns

Yanfeng Wang

Teng Yang

Bin Xu

Lianjie Qu

Yin Xia

Xiaoling Xu

Yuanxin Du

Lei Pan

Lu Fu

Xue Dong

Effendi Leobandung

Minghua Xuan

Kailash Gopalakrishnan

Yipeng Chen

Zhenhua Lv

Yu Wu

Hebin Zhao

Long Han

David James Frank

Malte Johannes Rasch

Ruiyong Wang

Chengjun Li

Junhong Yan

Jie Hu

Haisheng Wang

Yingming Liu

Shengji Yang

Peng Liu

Tao Wang

Silvia Melitta Mueller

Xibin Shao

Hailin Xue

Hui Zhang

Pan Li

Peng Wang

Hongjuan Liu

Naigang Wang

Shao Chun Li

Chun-Chen Yeh

Xiaolong Zhang

Tayfun Gokmen

Xiang Zhang

Chia-Yu Chen

Hui Xing Wang

Zhong Fang Yuan

Ji Li

Zhibin Ren

Ling Li

Tong Liu

Li Ni Zhang

Kerstin Claudia Schelm

Rong Fu

Youngseok Kim

Jungwook Choi

Jian Xu

Hong Yang

Huijuan Wang

Ankur Agrawal

Bingqiang Gui

Yanqing Chen

Leteng Weng

Yang You

Tian Ji Yang

Yu Tian

Hao Sheng

Runjia Huang

Zhiyuan Lin

Chang Rui Ren

Wenheng Guo

Zongzhuo Wu

Zhang Li

Jinfeng Li

Hongyan Li

Xuebo Zhang

Han Hu

Provided are techniques for identifying and correcting model focus drift during model training. A model is trained using an original dataset with data, and the is classified into clusters. A representation is assigned to each of the clusters. A first visualization dashboard with visualizations is generated, where each visualization represents a first data distribution of an associated cluster using the representation assigned to that cluster. The model is fine-tuned using a fine-tune dataset. A second visualization dashboard is generated by updating each visualization, where each visualization represents a second data distribution of the associated cluster. It is determined that a cluster of the clusters has focus drift based on changes between the first data distribution and the second data distribution. The focus drift is corrected by: adding the data of the cluster to the fine-tune dataset to form a combined dataset and fine-tuning the model using the combined dataset.

Identifying and correcting model focus drift during model training

Automatic container image registry selection is provided. A predefined number of top performing container image registries is selected from a top of a sorted list of a plurality of container image registries listed in descending order. A cluster of host nodes where a container is deployed is enabled to pull a same container image corresponding to the container utilizing different predefined pulling percentages from the predefined number of top performing container image registries. The cluster of host nodes utilizes a different predefined pulling percentage to pull the same container image from each respective top performing container image registry. A current real time average pulling speed of the cluster of host nodes for the same container image from each respective top performing container image registry of the predefined number of top performing container image registries is recorded over a defined time period.

Automatic container image registry selection

Disclosed are a hydroxypentyl benzoic acid diester compound, a method preparing the same, and an application of the same. This application uses a combination principle, on the one hand, the medicine concentration of butylphthalide in cerebral vessels is improved through a function of promoting the medicine to penetrate through a blood-brain barrier of dexborneol, and a multi-target multi-channel synergistic brain protection effect of the medicine is exerted; on the other hand, the treatment effect of dexborneol on cerebral ischemia is exerted. Anti-cerebral ischemia in-vivo pharmacodynamic activity evaluation is carried out on the synthesized compound, a high-activity candidate medicine is obtained, and industrial production can be achieved through the synthesis method.

Hydroxypentyl benzoic acid diester compound, and preparation method and pharmaceutical application thereof

Embodiments of this disclosure discloses a map data processing method, apparatus, computer device and storage medium, the method including: outputting navigation route information associated with a navigation user in a navigation interface of a map client application, the navigation interface including map layer information, the navigation route information being superimposed and displayed on the map layer information, and map field information of the map layer information being first field information; obtaining location information of the navigation user on the navigation route information; and updating and displaying the map field information from the first field information to second field information in the navigation interface when the location information is monitored to be in an intersection area on the navigation route information. By the embodiments of this disclosure, the safety during navigation may be improved.

Map data processing method and device, computer equipment, and storage medium

A scheme for modifying an image is disclosed, which includes receiving a source image having a first image configuration; determining a second image configuration for a target image; providing the received source image to an AI engine trained to identify, based on a set of rules related to visual features, candidate regions from the source image; generating proposal images based on the candidate regions, respectively; determining, based on prior aesthetical evaluation data, an aesthetical value of each regional proposal image; and selecting, based on the determined aesthetical value of each regional proposal image, one of the regional proposal images as the target image; extracting, from the AI engine, the target image; and causing the target image to be displayed via a display of a user device.

AI-based aesthetical image modification

An apparatus includes a memory and a processor coupled to the memory. The processor includes first and second sets of arithmetic units having first and second precision for floating-point computations, the second precision being lower than the first precision. The processor is configured to obtain a machine learning model trained in the first precision, to utilize the second set of arithmetic units to perform inference on input data, to utilize the first set of arithmetic units to generate feedback for updating parameters of the second set of arithmetic units based on the inference performed on the input data by the second set of arithmetic units, to tune parameters of the second set of arithmetic units based at least in part on the feedback generated by the first set of arithmetic units, and to utilize the second set of arithmetic units with the tuned parameters to generate inference results.

Mixed precision capable hardware for tuning a machine learning model

The present application provides a display panel, belonging to the field of display technology. The display panel includes a base substrate, and a first electro-conductive pattern and a second electro-conductive pattern which are arranged on the base substrate, wherein the first electro-conductive pattern includes a first electrode layer, the second electro-conductive pattern includes a second electrode layer, at least one of the first electro-conductive pattern and the second electro-conductive pattern further includes an antenna pattern, and the antenna pattern is insulated from the electrode layer in the electro-conductive pattern where the antenna pattern is located.

Display panel, and display apparatus and control method thereof

A method for multiple copies of a set of multi-kernel set operations in a hardware accelerated neural network includes a word line for receiving a pixel value of an input image. A bit line communicates a modified pixel value. An analog memory cell including a first capacitor stores a first kernel weight of a first kernel in one of a plurality of kernel sets such that the pixel value is operated on by the first kernel weight to produce the modified pixel value. A charge connection connects the first capacitor to at least a second capacitor storing a second kernel weight of a related kernel of a second one of the plurality of kernel sets such that charge is shared between the first capacitor and at least the second capacitor to normalize the first kernel weight and the second kernel weight.

Kernel sets normalization with capacitor charge sharing

A display assembly includes a display panel and at least one optical film group each disposed on a display surface. Each optical film group includes a quarter-wave plate, a reflective polarizer and an absorbing polarizer. The reflective polarizer includes a reflective portion capable of allowing light with a polarization direction parallel to a polarization axis of the reflective polarizer to pass through and reflecting light with a polarization direction perpendicular to the polarization axis. An orthographic projection of an effective light-emitting area of at least one sub-pixel is substantially within an orthographic projection of the reflective portion. The absorbing polarizer is capable of allowing light with a polarization direction parallel to a polarization axis of the absorbing polarizer to pass through and absorbing light with a polarization direction perpendicular to the polarization axis. The polarization axis of the reflective polarizer is parallel to the polarization axis of the absorbing polarizer.

Optical film group, display assembly and display device

Methods and systems of implementing a convolutional neural network are described. In an example, a structure may receive input signals and distribute the input signals to a plurality of unit cells. The structure may include a plurality of multi-kernel modules that may include a respective set of unit cells. A unit cell may correspond to an element of a kernel being implemented in the convolutional neural network and may include a storage component configured to store a weight of a corresponding element of the kernel. A first pass gate of the unit cell may be activated to pass a stored weight of the unit cell to a plurality of operation circuits in the corresponding unit cell, such that the stored weight may be applied to the input signals. The structure may generate a set of outputs based on the application of the stored weights to the input signals.

Multi-kernel configuration for convolutional neural networks

A computer-implemented method for training a random matrix network is presented. The method includes initializing a random matrix, inputting a plurality of first vectors into the random matrix, and outputting a plurality of second vectors from the random matrix to be fed back into the random matrix for training. The random matrix can include a plurality of two-terminal devices or a plurality of three-terminal devices or a film-based device.

Random matrix hardware for machine learning

A display apparatus having a plurality of subpixels is provided. The display apparatus includes an array substrate and a counter substrate facing the array substrate. The counter substrate includes a base substrate; an optical compensation device on the base substrate configured to adjust light emitting brightness values of the plurality of subpixels to target brightness values respectively; and a plurality of light shielding walls on the base substrate. The optical compensation device include a plurality of photosensors configured to respectively detect light emitting brightness values of the plurality of subpixels. A respective one of the plurality of light shielding walls is configured to at least partially shield a lateral side of a respective one of the plurality of photosensors from light emitted from adjacent subpixels.

Display apparatus, counter substrate of display apparatus, method of fabricating display apparatus

The present disclosure relates to an optical detection panel. The optical detection panel may include a first substrate and a second substrate opposite the first substrate, a photosensitive component and a driving thin film transistor at a side of the second substrate facing the first substrate, a first electrode and a second electrode at a side of the second substrate facing the first substrate, and a plurality of microlenses at a side of the photosensitive component opposite from the second substrate. The second electrode may be connected to the driving thin film transistor.

Optical detection panel, fabricating method thereof and display apparatus

The embodiments of the present disclosure provide a 3D display device and a manufacturing method thereof. The 3D display device includes a first substrate; a second substrate disposed opposite to the first substrate; a black matrix; and a grating. The black matrix and the grating are disposed on a side of the first substrate facing away from the second substrate; the black matrix and the grating are disposed in a same layer; and a side of the first substrate where the black matrix and the grating are located is a light exit side of the 3D display device.

3D device and manufacturing method thereof

A wireless charging assembly includes a substrate and a charging emission trace on the substrate. The wireless charging assembly further includes a touch structure on a side, which is away from the charging emission trace, of the substrate. A preparation method of the wireless charging assembly, a terminal device and a wireless charging method for the terminal device are also provided.

Wireless charging assembly and preparation method thereof, terminal device and wireless charging method thereof

Growing community of inventors

Ningbo, China

Xiao Lin Sun