Abstract: The present invention relates to adeno-associated viral (AAV) particles encoding siRNA molecules and methods for treating Huntington's Disease (HD).

Abstract: A method of constructing an adaptive multiply accumulate layer in a convolutional neural network, including determining an activation data map width, an activation data map height, a channel depth, a batch, a kernel width, a kernel height and a filter set number, setting a first dimension of an adaptive multiplier layer based on the activation data map width, setting a second dimension of the adaptive multiplier layer based on the channel depth, setting a third dimension of the adaptive multiplier layer based on the filter set number and constructing the adaptive multiplier layer based on the first dimension, the second dimension and the third dimension.

Abstract: A process for refining glycolide includes a step of extracting impurities from a crude glycolide with solvent A, and then a step of recrystallizing with solvent B. Solvent A includes at least two solvents and the solvents are miscible each other. In the process, two types of polar solvents are used for the refining of crude glycolide. The process has a good extraction effect on acidic impurities such as glycolic acid oligomers. The solvents are easily fully removed during the solid-liquid separation and the drying process. Glycolide obtained using this process may have a low free acid content.

Abstract: The present invention relates to convolution neural networks (CNN) and methods for improving computational efficiency of multiply accumulate (MAC) array structure Specifically, the invention relates to cutting of activation data into a number of tiles for increasing overall computation efficiency. The invention discloses techniques to cut an activation data into a plurality of tiles by using a 3-D convolution computation core and support bigger tensor sizes. Lastly, the invention provides adaptive scheduling of MAC array to achieve high utilization in multi-precision neural network acceleration.

Abstract: A method of constructing sub-graphs, includes receiving a directed acyclic graph (DAG), partitioning the directed acyclic graph into an at least one section, determining at least one hardware attribute, determining at least one DAG hardware limitation of the at least one section and determining a largest continuous node list of the at least one section in which the at least one hardware attribute meets the at least one DAG hardware limitation.

Abstract: A method of subgraph tile fusion in a convolutional neural network, including partitioning a network into at least one subgraph node, determining a layer order of at least one layer of the at least one subgraph node, determining a input layer of the at least one subgraph node, determining a weight layer of the at least one subgraph node, determining a output layer of the at least one subgraph node and fusing the at least one subgraph node, the input layer, the weight layer and the output layer in the layer order.

Abstract: A method of hierarchical structuring a multi-level memory in a convolutional neural network, includes partitioning a memory into a plurality of sections, partitioning the plurality of sections into a plurality of stripes, utilizing input data from the plurality of stripes in a MAC array, outputting an intermediate result from the MAC array to at least one of the plurality of stripes of a result buffer, looping back the intermediate result from the at least one of the plurality of stripes of the result buffer to at least one of the plurality of stripes of an input data buffer and outputting a final result from the at least one of the plurality of stripes of the result buffer to at least one of the plurality of stripes of an output buffer.

Abstract: A generalized application virtualization platform and method are provided for business use on the web and a mini server. The method includes: (1) Initializing, application monitoring, registration and login; (2) Accessing to a server via Access protocol from a client-side SOD (software on-demand) controller; (3) Content selecting via keyword search, with the selected contents provided by a content management protocol; (4) Choosing one of the available application virtualization methods according to bandwidth needed by the user and the resources consumed in a data center; (5) Binding of the executable software for the application virtualization method and the contents selected, and running the executable software; and (6) Metering and billing: Meter the start and end time of the execution of application virtualization method and calculate the billing.

Abstract: A display panel and a method of manufacturing thereof are provided. The pixel defining block is formed by one-time exposing the column spacer column and the bank with a halftone mask, thereby omitting the process of individually forming the bank, and manufacturing time and manufacturing cost are reduced. The organic functional layer disposed above the bank is not directly contacted with the first electrode due to the bank structure, and the bank causes the defined region to be disconnected from the first electrode, and no current flows through the defined region. Thus, the defined region does not emit light, and the pixel region emits uniform light. The material of the bank is an organic photoresist, so hydrogen and oxygen are not introduced into the bank, and therefore not affect the thin film transistor, and improves stability of the display panel.

Abstract: A display panel and a method of manufacturing thereof are provided. The pixel defining block is formed by one-time exposing the column spacer column and the bank with a halftone mask, thereby omitting the process of individually forming the bank, and manufacturing time and manufacturing cost are reduced. The organic functional layer disposed above the bank is not directly contacted with the first electrode due to the bank structure, and the bank causes the defined region to be disconnected from the first electrode, and no current flows through the defined region. Thus, the defined region does not emit light, and the pixel region emits uniform light. The material of the bank is an organic photoresist, so hydrogen and oxygen are not introduced into the bank, and therefore not affect the thin film transistor, and improves stability of the display panel.

Abstract: A method of constructing an adaptive multiply accumulate layer in a convolutional neural network, including determining an activation data map width, an activation data map height, a channel depth, a batch, a kernel width, a kernel height and a filter set number, setting a first dimension of an adaptive multiplier layer based on the activation data map width, setting a second dimension of the adaptive multiplier layer based on the channel depth, setting a third dimension of the adaptive multiplier layer based on the filter set number and constructing the adaptive multiplier layer based on the first dimension, the second dimension and the third dimension.

Abstract: A method of hierarchical structuring a multi-level memory in a convolutional neural network, includes partitioning a memory into a plurality of sections, partitioning the plurality of sections into a plurality of stripes, utilizing input data from the plurality of stripes in a MAC array, outputting an intermediate result from the MAC array to at least one of the plurality of stripes of a result buffer, looping back the intermediate result from the at least one of the plurality of stripes of the result buffer to at least one of the plurality of stripes of an input data buffer and outputting a final result from the at least one of the plurality of stripes of the result buffer to at least one of the plurality of stripes of an output buffer.

Abstract: A method of subgraph tile fusion in a convolutional neural network, including partitioning a network into at least one subgraph node, determining a layer order of at least one layer of the at least one subgraph node, determining a input layer of the at least one subgraph node, determining a weight layer of the at least one subgraph node, determining a output layer of the at least one subgraph node and fusing the at least one subgraph node, the input layer, the weight layer and the output layer in the layer order.

Abstract: Techniques for synchronizing audio and video data being transferred from one device to another device are disclosed. According to one aspect of the present invention, a periodic signal is generated in a host device (e.g., a personal computer) and transferred to a thin device (e.g., a PC camera) via an interface (e.g., a USB). The periodic signal is used to adjust local clocks in the thin device to generate audio and video data. As a result, the audio and video data is synchronized. According to another aspect of the present invention, an error detection unit and a correction unit are provided to ensure that there is no error in the periodic signal being transferred from the host device. If an error is detected, a patch to the signal is generated.

Abstract: The invention issued is a generalized application virtualization method for business use on the web and the mini server, including: (1) Initializing (which includes obtaining parameters, reading configuration file, and initializing a DDE service), application monitoring, registration and login. (2) Accessing to a server via Access Protocol from a client-side SOD controller (aka SOD client controller, or SOD controller). (3) Content selecting via keyword search, with the selected contents provided by the Content Management Protocol. (4) Choosing one of the available application virtualization methods according to bandwidth needed by the user and the resources consumed in the data center. (5) Binding of the executable software for the application virtualization method and the contents selected, and running of that executable software. (6) Metering and billing: Meter the start and end time of the execution of application virtualization method and calculate the billing.

Abstract: Techniques for increasing the efficiency in se of a data bus are disclosed. A bus system includes a bus for transferring data, and a device, being coupled to the bus for data transfer, including a functional module providing data required to be transferred over the bus and a first-in-first-out device (FIFO) coupled between the functional module and the bus. The FIFO is provided to store pending data from the functional module before the bus is granted to the functional portion in responding to a bus request (BUS_REQ) for transferring the data over the bus, wherein topside data in the FIFO always drives the bus so that the topside data is immediately transferred over the bus once the bus is granted to the device or the functional module.