Abstract: An algorithm for a sparse attention operation is provided, and the algorithm includes following steps. An attention probability matrix is calculated based on a query matrix and a key matrix. A pruning ratio is calculated based on the attention probability matrix. The attention probability matrix is pruned based on the pruning ratio to obtain a pruned attention probability matrix. A value matrix is pruned based on the pruning ratio to obtain a pruned value matrix. The pruned attention probability matrix and the pruned value matrix are multiplied to obtain an attention matrix.

Abstract: The present invention relates to the use of a class of 1,4-dihydro-naphthyridine derivatives in the treatment of tumors.

Abstract: A processing element architecture adapted to a convolution comprises a plurality of processing elements and a delayed queue circuit. The plurality of processing elements includes a first processing element and a second processing element, wherein the first processing element and the second processing element perform the convolution according to a shared datum at least. The delayed queue circuit connects to the first processing element and connects to the second processing element. The delayed queue circuit receives the shared datum sent by the first processing element, and sends the shared datum to the second processing element after receiving the shared datum and waiting for a time interval.

Abstract: A system and method for providing video conferencing are described herein. Original video streams from one or more cameras are sampled to form downscaled versions of the video stream. The downscaled version is analyzed to find regions of interest and generate metadata describing aspects of the regions of interest, including a region that includes a preferred speaker and the best view of each participant present together in the video conference. Cropping instructions are generated from the metadata, and portions of the original video stream are removed to form a set of presentation views for each participant, including the preferred speaker. The set of presentation views is stitched together to form a composite display for each participant viewing the stream remotely.

Abstract: A system and method for providing an integrated system of endpoints are described, even though the endpoints are distributed over several locations. A distributed controller operates to link the endpoints together and provide a unified view of endpoints to a conference provider. In addition, the distributed controller can assign the workload of some endpoints to other endpoints that are better able to handle the workload. The distributed controller is implemented by cooperating processes in the endpoints.

Abstract: An inflatable product includes an inflatable chamber and an elongated element. The inflatable chamber includes fusion portions. The elongated element is flexible, is disposed in the inflatable chamber, includes enlarged portions, and is connected to the inflatable chamber with the enlarged portions constrained by the fusion portions. The elongated element is in tension for controlling expansion of the inflatable chamber to a desired shape when the inflatable chamber is inflated to expand. The enlarged portions are constrained by the fusion portions to avoid separation of the elongated element from the inflatable chamber when the inflatable chamber is inflated to expand.

Abstract: A system and method for providing an integrated system of endpoints are described, even though the endpoints are distributed over several locations. A distributed controller operates to link the endpoints together and provide a unified view of endpoints to a conference provider. In addition, the distributed controller can assign the workload of some endpoints to other endpoints that are better able to handle the workload. The distributed controller is implemented by cooperating processes in the endpoints.

Abstract: A system and method for providing video conferencing are described herein. Original video streams from one or more cameras are sampled to form downscaled versions of the video stream. The downscaled version is analyzed to find regions of interest and generate metadata describing aspects of the regions of interest, including a region that includes a preferred speaker and the best view of each participant present together in the video conference. Cropping instructions are generated from the metadata, and portions of the original video stream are removed to form a set of presentation views for each participant, including the preferred speaker. The set of presentation views is stitched together to form a composite display for each participant viewing the stream remotely.

Abstract: Embodiments of the disclosure provided herein generally relate to methods and video system components that have integrated background differentiation capabilities that allow for background replacement and/or background modification. In some embodiments, undesired portions of video data generated in a video environment are separated from desired portions of the video data by taking advantage of the illumination and decay of the intensity of electromagnetic radiation, provided from an illuminator, over a distance. Due to the decay of intensity with distance, the electromagnetic radiation reflected from the undesired background has a lower intensity when received by the sensor than the electromagnetic radiation reflected from the desired foreground. The difference in the detected intensity at the one or more wavelengths can then be used to separate and/or modify the undesired background from the desired foreground for use in a video feed.

Abstract: Various solutions for global navigation satellite system (GNSS) operations with respect to user equipment and network node in mobile communications are described. An apparatus may transmit a GNSS position fix time duration and a GNSS validity duration to a network node during an initial access. The apparatus may receive a configuration based on the GNSS position fix time duration and the GNSS validity duration from the network node in a connected mode. The apparatus may perform a GNSS measurement based on the configuration. The apparatus may transmit a new GNSS validity duration to the network node in the connected mode in an event that the apparatus completes the GNSS measurement.

Abstract: A posture stability detection system including a pressure pad, a camera and a processor is provided. The pressure pad is configured to sense a pressure distribution exerted by a to-be-tested subject. The camera is configured to capture a plurality of posture images of the to-be-tested subject. The processor is configured to: obtain a COP position of the to-be-tested subject according to the pressure distribution in a sample; obtain a COP distribution circle of the to-be-tested subject according to the COP positions in several samples; obtain a COP stability value of the to-be-tested subject according to the COP distribution circles; analyze each posture image to obtain a backbone stability value; and, obtain a posture stability value according to the backbone stability value and the COP stability value.

Abstract: An electronic component carrier board includes a plurality of carrier plates, a plurality of insulating layers and a plurality of conductive members. The carrier plates are formed with a plurality of slots. The insulating layers are disposed between the carrier plates. The conductive members are made of conductive material and each is disposed in one of the slots. Each of the conductive members includes a tubular body that is inserted in a corresponding one of the slots and that defines a conductive space, and a pin body that is inserted in the conductive space and that is electrically connected with the tubular body. For each of the conductive members, at least one of the tubular body and the pin body is detachably disposed in the corresponding one of the slots.

Abstract: A hardware-aware zero-cost neural network architecture search system is configured to perform the following. A neural network search space is divided into multiple search blocks. Each of the search blocks includes multiple candidate blocks. The candidate blocks are guided and scored through a latent pattern generator. The candidate blocks in each of the search blocks are scored through a zero-cost accuracy proxy. One of the candidate blocks included in each of the search blocks is sequentially selected as selected candidate blocks, the selected candidate blocks are combined into multiple neural networks to be evaluated, and network potential of the neural networks to be evaluated is calculated according to scores of the selected candidate blocks. One neural network to be evaluated with the highest network potential is selected to determine the corresponding selected candidate blocks.

Abstract: An artificial intelligence accelerator includes an external command dispatcher, a first data access unit, a second data access unit, a global buffer, an internal command dispatcher, and a data/command switch. The external command dispatcher receives an address and access information. The external command dispatcher sends the access information to one of the first data access unit and the second data access unit, the first data access unit receives first data from a storage device according to the access information, and sends the first data to the global buffer. The second data access unit receives second data from the storage device according to the access information, and sends the second data. The data/command switch receives the address and the second data from the second data access unit, and sends the second data to one of the global buffer and the internal command dispatcher.

Abstract: An inflatable product includes an inflatable chamber and a supplemental layer. The supplemental layer is disposed on the inflatable chamber. Further, the supplemental layer is directly or indirectly fixed to the inflatable chamber by sewing. The inflatable chamber is made of plastic. The material of the supplemental layer is different from that of the inflatable chamber.

Abstract: Disclosed is the use of a composition containing cilostazol or a pharmaceutically acceptable salt thereof and edaravone in the preparation of a drug for treating cerebrovascular diseases, especially ischemic cerebrovascular diseases. The experiment result shows that the tail vein administration of 1-15 mg/kg of cilostazol or 1.67-8.33 mg/kg of edaravone on rats with a focal cerebral ischemia-reperfusion injury can significantly ameliorate neurological defects in rats with an MCAO and reduce a cerebral infarction area; and performing compound combination within the dosage range (the mass ratio of cilostazol:edaravone is 1:5 to 5:1) can be synergistic. The tail vein administration of 3.33-16.67 mg/kg of cilostazol or 3.33-16.67 mg/kg of edaravone on mice with a focal cerebral ischemia-reperfusion injury can significantly ameliorate neurological defects in mice with an MCAO and reduce a cerebral infarction area.

Abstract: The present invention belongs to the pharmaceutical field, and relates to an application of a cilostazol and d-borneol composition in preparing a drug for treating cerebrovascular disease, in particular ischemic cerebrovascular disease. The present invention specifically relates to a composition containing cilostazol or a pharmaceutically acceptable salt thereof and d-borneol, and an application of the composition in preparing a drug for treating cerebrovascular disease, in particular ischemic cerebrovascular disease.

Abstract: Apparatus and methods are provided for enhanced AM RLC for NB-IoT UE in the NTN. In one novel aspect, the UE is configured with HARQ disabled, and configured with fast polling and/or fast status report for AM RLC enhancement. In one embodiment, the UE is configured with shortened AM window size, and wherein the shortened window size is smaller than an AM window size for an NB-IoT device. In other embodiments, the fast polling is triggered based on PDU threshold or byte threshold, or triggered based on an NTN polling timer, or triggered based on a combination of an NTN polling timer and one or more threshold triggers. In one embodiment, UE is configured to autonomously send a status report when a reception failure of an RLC PDU is detected, and wherein the status report is sent by a semi-persistent scheduling (SPS) uplink resource.

Abstract: A system and method for providing video conferencing are described herein. Original video streams from one or more cameras are sampled to form downscaled versions of the video stream. The downscaled version is analyzed to find regions of interest and generate metadata describing aspects of the regions of interest, including a region that includes a preferred speaker and the best view of each participant present together in the video conference. Cropping instructions are generated from the metadata, and portions of the original video stream are removed to form a set of presentation views for each participant, including the preferred speaker. The set of presentation views is stitched together to form a composite display for each participant viewing the stream remotely.

Abstract: Various examples and schemes pertaining to enabling or disabling hybrid automatic repeat request (HARQ) feedback for multi-transport block (multi-TB) in an Internet-of-Things (IoT) non-terrestrial network (NTN) are described. A user equipment (UE) receives a configuration (e.g., from a network). Then, the UE applies the configuration to disable or enable a HARQ feedback regarding multiple TBs.