Abstract: A device receives an image. The image is applied to a sequence of encoders that includes a first encoder and one or more subsequent encoders, each including a U network encoder which performs convolution neural network processing. The one or more subsequent encoders receive a down-sampled version of the image from a prior encoder. The sequence of encoders form a first dimension orthogonal to the orthogonal dimension. A result of the sequence of encoders is applied to a sequence of decoders that includes a first decoder and one or more subsequent decoders. A respective decoder comprises a U network decoder that performs convolution neural network processing of an up-sampled version of the image from the first decoder. Probability outputs are produced from paired encoders and decoders in the sequence of encoders and the sequence of decoders. The probability outputs are combined to form a final output.

Abstract: A power module may include a plurality of dies and a heat spreader. The plurality of dies may include a first die subset and a second die subset. The first die subset may be associated with a power module hot spot, such that the first die subset may be associated with a higher operating temperature than the second die subset. The heat spreader may underlie each die of the first die subset. The heat spreader may be operable to reduce the operating temperature of the first die subset.

Abstract: Embodiments of the present disclosure relate to the field of electronic device manufacturing, and in particular, to multi-layered epitaxial stacks, such as complementary field-effect-transistors (cFETs). A method is used to fabricate a layered middle dielectric isolation (MDI) structure and carbon-doping of epitaxially grown silicon germanium layers together in the cFETs. In some embodiments, by integrating the layered MDI structure together with carbon-doping of SiGe layers into the cFETs, relaxation, wafer bow, and defects in a stack have been significantly reduced when compared to traditional stacks. Advantageously, multi-layered epitaxial stacks incorporate a greater number of silicon channels (e.g., pMOS and nMOS channels) when compared to traditional stacks. Furthermore, the selectivity in the downstream processes is improved by an order of magnitude.

Abstract: Embodiments of the present disclosure relate to the field of electronic device manufacturing, and in particular, to multi-layered epitaxial stacks, such as complementary field-effect-transistors (cFETs). A method is used to fabricate a layered middle dielectric isolation (MDI) structure and carbon-doping of epitaxially grown silicon germanium layers together in the cFETs. In some embodiments, by integrating the layered MDI structure together with carbon-doping of SiGe layers into the cFETs, relaxation, wafer bow, and defects in a stack have been significantly reduced when compared to traditional stacks. Advantageously, multi-layered epitaxial stacks incorporate a greater number of silicon channels (e.g., pMOS and nMOS channels) when compared to traditional stacks. Furthermore, the selectivity in the downstream processes is improved by an order of magnitude.

Abstract: Systems and methods for computer-assisted design of an inductor are described. Target specifications for an inductor are received. An inductor design is generated segment-by-segment using a reinforcement learning agent to generate segment parameters for each added segment. The reinforcement learning agent implements a policy that is learned using a reward computed based on performance of the generated inductor design relative to the target specifications. The generated inductor design is outputted as a candidate inductor design after determining that the generated inductor design satisfies a predefined performance threshold.

Abstract: A device receives an image. The image is applied to a sequence of encoders that includes a first encoder and one or more subsequent encoders, each including a U network encoder which performs convolution neural network processing. The one or more subsequent encoders receive a down-sampled version of the image from a prior encoder. The sequence of encoders form a first dimension orthogonal to the orthogonal dimension. A result of the sequence of encoders is applied to a sequence of decoders that includes a first decoder and one or more subsequent decoders. A respective decoder comprises a U network decoder that performs convolution neural network processing of an up-sampled version of the image from the first decoder. Probability outputs are produced from paired encoders and decoders in the sequence of encoders and the sequence of decoders. The probability outputs are combined to form a final output.

Abstract: Multi-chip module packaging technologies for GaN and other devices are described. The power module packaging technology described can be applied to all types of medium-voltage devices, such as silicon (Si), silicon carbide (SiC), gallium nitride (GaN), or the latest gallium oxide (Ga2O3) devices. In one example, a power module includes a first substrate, a second substrate, a sintered-silver semiconductor die pillar, the pillar being positioned between the first substrate and the second substrate, a terminal on a first side of the power module, and a terminal on a second side of the power module.

Abstract: A device receives an image. The image is applied to a sequence of encoders that includes a first encoder and one or more subsequent encoders, each including a U network encoder which performs convolution neural network processing. The one or more subsequent encoders receive a down-sampled version of the image from a prior encoder. The sequence of encoders form a first dimension orthogonal to the orthogonal dimension. A result of the sequence of encoders is applied to a sequence of decoders that includes a first decoder and one or more subsequent decoders. A respective decoder comprises a U network decoder that performs convolution neural network processing of an up-sampled version of the image from the first decoder. Probability outputs are produced from paired encoders and decoders in the sequence of encoders and the sequence of decoders. The probability outputs are combined to form a final output.

Abstract: The present disclosure provides a design method for an active disturbance rejection roll controller of a vehicle under disturbance of complex sea conditions, including: step 1: establishing a vehicle roll attitude control model; step 2: designing an active disturbance rejection controller (ADRC) on the basis of the control model in step 1 and a pole placement method; and step 3: performing an active disturbance rejection roll control by using the active disturbance rejection controller in step 2. The present disclosure solves the problem of a stable control of the vehicle under the disturbance of the complex sea conditions.

Abstract: A non-transitory computer readable storage medium has instructions executed by a processor to receive an ultrasound image. The ultrasound image is applied to a sequence of encoders where each encoder in the sequence of encoders performs convolution neural network processing of a down-sampled version of the ultrasound image from a prior encoder, the sequence of encoders form a first dimension. The ultrasound image is applied to a transition encoder with an orthogonal dimension to the first dimension. The ultrasound image is applied to a sequence of decoders where each decoder in the sequence of decoders performs convolution neural network processing of an up-sampled version of the ultrasound image from a prior decoder, the sequence of decoders form a second parallel dimension to the first dimension. Encoder and decoder configurations and the first dimension, the orthogonal dimension and the second parallel dimension thereby define a nested U network architecture.

Abstract: The present disclosure provides a design method for an active disturbance rejection roll controller of a vehicle under disturbance of complex sea conditions, including: step 1: establishing a vehicle roll attitude control model; step 2: designing an active disturbance rejection controller (ADRC) on the basis of the control model in step 1 and a pole placement method; and step 3: performing an active disturbance rejection roll control by using the active disturbance rejection controller in step 2. The present disclosure solves the problem of a stable control of the vehicle under the disturbance of the complex sea conditions.

Abstract: Systems and methods for computer-assisted design of an inductor are described. Target specifications for an inductor are received. An inductor design is generated segment-by-segment using a reinforcement learning agent to generate segment parameters for each added segment. The reinforcement learning agent implements a policy that is learned using a reward computed based on performance of the generated inductor design relative to the target specifications. The generated inductor design is outputted as a candidate inductor design after determining that the generated inductor design satisfies a predefined performance threshold.

Abstract: A non-transitory computer readable storage medium has instructions executed by a processor to receive an ultrasound image. The ultrasound image is applied to a sequence of encoders where each encoder in the sequence of encoders performs convolution neural network processing of a down-sampled version of the ultrasound image from a prior encoder, the sequence of encoders form a first dimension. The ultrasound image is applied to a transition encoder with an orthogonal dimension to the first dimension. The ultrasound image is applied to a sequence of decoders where each decoder in the sequence of decoders performs convolution neural network processing of an up-sampled version of the ultrasound image from a prior decoder, the sequence of decoders form a second parallel dimension to the first dimension. Encoder and decoder configurations and the first dimension, the orthogonal dimension and the second parallel dimension thereby define a nested U network architecture.

Abstract: The present invention discloses a vertical structure nonpolar LED chip on a lithium gallate substrate and a preparation method therefor. According to the method, LED epitaxial wafers are grown on a lithium gallate substrate, wherein the LED epitaxial wafers comprise a GaN buffer layer grown on the lithium gallate substrate, a non-doped GaN layer on the GaN buffer layer, an n-type doped GaN thin film on the non-doped GaN layer, an InGaN/GaN quantum well on the n-type doped GaN thin film and a p-type doped GaN thin film on the InGaN/GaN quantum well. Then, electrode patterns are prepared on the surfaces of the LED epitaxial wafers by the steps of spin coating, photoetching, developing and cleaning, and an electrode metal is sequentially deposited on the upper surfaces of the epitaxial wafers. Then, the LED epitaxial wafers are transferred to a copper substrate.

Abstract: A method and an apparatus for transmitting frame data between a near-end device and a remote device are provided. The method includes: generating, by the near-end device, a base frame in a user-defined frame format, wherein a PHY converting chip is built in the near-end device, the base frame includes a first number of super groups, each super group includes a second number of base groups, and each base group includes media access control (MAC) frame structure data and an interframe gap; matching duration of the MAC frame structure data and the interframe gap with an output timing sequence of the PHY converting chip; and converting the base frame into an optical fiber signal through the PHY converting chip, and sending the optical fiber signal to the remote device.

Abstract: The present invention discloses a vertical structure nonpolar LED chip on a lithium gallate substrate and a preparation method therefor. According to the method, LED epitaxial wafers are grown on a lithium gallate substrate, wherein the LED epitaxial wafers comprise a GaN buffer layer grown on the lithium gallate substrate, a non-doped GaN layer on the GaN buffer layer, an n-type doped GaN thin film on the non-doped GaN layer, an InGaN/GaN quantum well on the n-type doped GaN thin film and a p-type doped GaN thin film on the InGaN/GaN quantum well. Then, electrode patterns are prepared on the surfaces of the LED epitaxial wafers by the steps of spin coating, photoetching, developing and cleaning, and an electrode metal is sequentially deposited on the upper surfaces of the epitaxial wafers. Then, the LED epitaxial wafers are transferred to a copper substrate.

Abstract: A method and an apparatus for transmitting frame data between a near-end device and a remote device are provided. The method includes: generating, by the near-end device, a base frame in a user-defined frame format, wherein a PHY converting chip is built in the near-end device, the base frame includes a first number of super groups, each super group includes a second number of base groups, and each base group includes media access control (MAC) frame structure data and an interframe gap; matching duration of the MAC frame structure data and the interframe gap with an output timing sequence of the PHY converting chip; and converting the base frame into an optical fiber signal through the PHY converting chip, and sending the optical fiber signal to the remote device.