Abstract: A tensor accelerator includes two tile execution units and a bidirectional queue. Each of the tile execution units includes a buffer, a plurality of arithmetic logic units, a network, and a selector. The buffer includes a plurality of memory cells. The network is coupled to the plurality of memory cells. The selector is coupled to the network and the plurality of arithmetic logic units. The bidirectional queue is coupled between the selectors of the tile execution units.

Abstract: A method for training a locomotion controller of a robotic animal comprising obtaining a motion data of a reference animal, an environmental parameter, and a disturbance parameter; generating a plurality of primitive distributions and a first primitive influence according to the motion data by a policy network; selecting a current state of the reference animal from the motion data and set an adapting state of the reference animal; generating a second primitive influence, by the policy network, according to the current state, the adapting state, and the plurality of primitive distributions at least; and training the policy network according to a result determined by a discriminator according to the first primitive influence and the second primitive influence.

Abstract: Embodiments of the disclosure provide for improved print position compensation, for example to improve accuracy of print job(s) performed by a printer. The print position compensation enables an offset of the time until printing occurs on a print media to account for changes and/or erroneous movement in a print media, such as due to slippage and/or other results of a force applied to the print media. Particular embodiments determine data values derived both for an output phase and a retraction phase of the printer's operation. Various embodiments generate a print position compensation based on sensor-based edge position distances determined during each of a media output phase and a media retraction phase. Alternatively or additionally various embodiments generate a print position compensation based on sensor-based media movement phase timestamp differentials determined during each of a media output phase and a media retraction phase.

Abstract: An LED device includes a substrate, a conductive layer, an LED chip, and a discharge element. The substrate has upper and lower surfaces and four edges interconnected to one another and surrounding the upper surface. The conductive layer is formed on the upper surface, and has first and second regions electrically separated by a trench. The trench has a first segment inclined relative to each edge of the substrate by a predetermined angle ranging between 0 and 90 degrees, and a second segment connected to the first segment. The LED chip is disposed across the first segment, and the discharge element is disposed across the second segment, both interconnecting the first and second regions.

Abstract: Double patterning techniques described herein may reduce corner rounding, etch loading, and/or other defects that might otherwise arise during formation of a deep trench isolation (DTI) structure in a pixel array. The double patterning techniques include forming a first set of trenches in a first direction and forming a second set of trenches in a second direction in a plurality of patterning operations such that minimal to no etch loading and/or corner rounding is present at and/or near the intersections of the first set of trenches and the second set of trenches.

Abstract: Implantation mask formation techniques described herein include increasing an initial aspect ratio of a pattern in an implantation mask by non-lithography techniques, which may include forming a resist hardening layer on the implantation mask. The pattern may be formed by photolithography techniques to the initial aspect ratio that reduces or minimizes the likelihood of pattern collapse during formation of the pattern. Then, the resist hardening layer is formed on the implantation mask to increase the height of the pattern and reduce the width of the pattern, which increases the aspect ratio between the height of the openings or trenches and the width of the openings or trenches of the pattern. In this way, the pattern in the implantation mask may be formed to an ultra-high aspect ratio in a manner that reduces or minimizes the likelihood of pattern collapse during formation of the pattern.

Abstract: A wound treatment system includes a wound dressing, a negative-pressure source, a light-emitting device, an image recognition module, and a control module. The wound dressing includes a housing with an opening, and a barrier layer with a conduit pathway separating the inside of the housing from the opening to form a receiving space. The negative-pressure source is fluidly connected to the wound dressing to remove tissue exudate of the wound. The light-emitting device is disposed in the receiving space, optically coupled to the barrier layer and can emit light of different wavelength bands as the detection light required by the image recognition module disposed on a side of the opening of the housing of the wound dressing and provide specific wavelengths of phototherapy required by different stages of the wound.

Abstract: A method includes forming a first photoresist layer on a front side of a device substrate and having first trenches spaced apart from each other. A first implantation process is performed using the first photoresist layer as a mask to form first isolation regions in the device substrate. A second photoresist layer is formed on the front side and has second trenches. A second implantation process is performed using the second photoresist layer as a mask to form second isolation regions in the device substrate and crossing over the first isolation regions. A third photoresist layer is formed on the front side and has third trenches spaced apart from each other. A third implantation process is performed using the third photoresist layer as a mask to form third isolation regions in the device substrate and crossing over the first isolation regions but spaced apart from the second isolation regions.

Abstract: Examples of the present disclosure relate generally to a printing apparatus and, more particularly, to apparatuses, systems, and methods for printing utilizing laser print head and reactive media.

Abstract: A vortex fountain or waterer that incorporates measures to induce swirl in liquid moving toward a drain opening of the fountain to inhibit the accumulation of debris is provided. The fountain may have a bowl including a base surface having a slope that is configured to create a vortex in the liquid flowing toward the drain opening to induce movement of a quantity of water, where the bowl is located above a basin configured to hold the liquid. The fountain may also include a pump outlet jet that can be positioned about the bowl to induce swirl about the bowl. The fountain may also include a drain tube located beneath the drain opening, and an extended guide tube extending downwardly towards the bottom of the basin. When the liquid passes through the drain, drain tube, and guide tube, oxygen is incorporated into the liquid as the liquid reaches the basin.

Abstract: A light-emitting diode (LED) device includes a substrate, an electrically conductive layer, a first LED chip, and an anti-electrostatic discharge element. The substrate has opposite upper and lower surfaces. The electrically conductive layer is formed on the upper surface of the substrate, and has first and second regions that are electrically separated from each other by a trench structure. The trench structure has a first segment and a second segment which connects and is not collinear with the first segment. The first LED chip is disposed across the first segment, and the anti-electrostatic discharge element is disposed across the second segment, both interconnecting the first and second regions.

Abstract: A method for generating a loop video comprises: obtaining an input video including multiple frames with a first frame, with each of frames having multiple pixels; extracting a moving object corresponding to a moving pixel region in the first frame with the moving pixel region having at least two pixels; inputting multiple candidate periods to a target function respectively to calculate multiple errors of the moving pixel region; determining a start frame and a loop period of the moving pixel region for each error, with the loop period being associated with one of the candidate periods; generating multiple output frames according to start frames and loop periods; and generating an output frame sequence from the output frames according to a loop parameter, wherein the output frame sequence corresponds to the loop video.

Abstract: A method of transmitting and merging data is adapted to a sender and a receiver that are in communication with each other. The method comprises a sending stage and a receiving stage. The sending stage comprises: transmitting a first block data, a second block data and a third block data to the receiver by the sender; obtaining a fourth block data and a fifth block data by the sender; and transmitting the third, fourth and fifth block data to the receiver by the sender. The receiving stage comprises: receiving the first, second, and third block data by the receiver; merging the first, second and third block data to perform a convolution operation by the receiver; receiving the third, fourth and fifth block data by the sender; and merging the third, fourth and fifth block data to perform another convolution operation.

Abstract: A method for generating reconstruction a reconstructed image is adapted to an input image having a target object. The method comprises converting the input image into a feature map with vectors by an encoder; performing a training procedure according to training images of reference objects to generate feature prototypes associated with the training images and store the feature prototypes to a memory; selecting a part of feature prototypes from the feature prototypes stored in the memory according to similarities between the feature prototypes and the feature vectors; generating a similar feature map according the part of feature prototypes and weights, wherein the weights represents similarities between the part of feature prototypes and the feature vectors; and converting the similar feature map into the reconstructed image by a decoder; wherein the encoder, the decoder and the memory form an auto-encoder.

Abstract: The laser device includes a substrate, a laser element disposed on the substrate for emitting a laser light ray, a light guide member disposed on the substrate, and a wavelength conversion layer. The light guide member is light-transmissible and thermally conductive, and has at least one reflection surface for reflecting the laser light ray from the laser element so as to change travelling direction of the laser light ray. The wavelength conversion layer converts wavelength of the laser light ray from the light guide member to result in a laser beam, and contacts the light guide member so that heat from the wavelength conversion layer is transferred to the substrate through the light guide member.

Abstract: Provided are a switching device with an auxiliary power unit and a wireless receiver function, and a lamp system using the same, wherein the switching device is connectable to a lamp and includes a power switching portion, which transmits external electricity to the lamp when in a power supplying mode and stops doing so when in a power-off mode; and the auxiliary power unit in the switching device keeps powering other components of the switching device, so while the power switching portion is in the power-off mode, a wireless module in the switching device can still transmit a wireless signal from a wireless controller to a control unit to adjust the power switching portion to the power supplying mode. Thus, even when the switching device is in a power-off state, a user can control the lamp by sending a wireless control signal to the switching device through the wireless controller.

Abstract: A method for training a defect detector comprises: obtaining a first reference image of a first reference object, wherein the first reference object has a defect and the first reference image has a first label indicating the defect; training a reconstruction model according to a second reference image of a second reference object associated with the first reference object, wherein a defect level of the second reference object is in a tolerable range with an upper limit; obtaining a target image of a target object associated with the first reference object and the second reference object; generating a second label according to the target image, the reconstruction model and an error calculation procedure, wherein the second label comprises a defect of the target object; and training a defect detector by performing a machine learning algorithm according to the first reference image, the target image and the second label.

Abstract: A log processing device and a log processing method thereof are provided. The log processing device divides the original log data into a plurality of block data, transforms a numeric variable of each of the block data into a representative code, and determines whether to perform a combination process for continuous block data to generate a plurality of combinational block data according to a data integrity of each of the block data. The log processing data takes the combinational block data as a log template, and each of the combinational block data corresponds to an event.

Abstract: The present invention provides a white light LED package structure and a white light source system, which includes a substrate, an LED chip, and a wavelength conversion material layer. The peak emission wavelength of the LED chip is between 400 nm and 425 nm; the peak emission wavelength of the wavelength conversion material layer is between 440 nm and 700 nm, and the wavelength conversion material layer absorbs light emitted from the LED chip and emits a white light source; and the emission spectrum of the white light source is set as P(?), the emission spectrum of a blackbody radiation having the same color temperature as the white light source is S(?), P(?max) is the maximum light intensity within 380-780 nm, S(?max) is the maximum light intensity of the blackbody radiation within 380-780 nm, D(?) is a difference between the spectrum of the white light LED and the spectrum of the blackbody radiation, and within 510-610 nm, the white light source satisfies: D(?)=P(?)/P(?max)?S(?)/S(?max), ?0.

Abstract: A method for generating a loop video comprises: obtaining an input video including multiple frames with a first frame, with each of frames having multiple pixels; extracting a moving object corresponding to a moving pixel region in the first frame with the moving pixel region having at least two pixels; inputting multiple candidate periods to a target function respectively to calculate multiple errors of the moving pixel region; determining a start frame and a loop period of the moving pixel region for each error, with the loop period being associated with one of the candidate periods; generating multiple output frames according to start frames and loop periods; and generating an output frame sequence from the output frames according to a loop parameter, wherein the output frame sequence corresponds to the loop video.