Abstract: A method, computer system, and a computer program product for smart SDN is provided. The present invention may include recording and clustering a pod's behavior to generate a behavior transition model for the pod. The present invention may include watching a behavior of the pod and comparing the behavior to the generated behavior transition model. The present invention may include triggering a network policy change based on determining that the behavior of the pod is a misbehavior.

Abstract: The present invention discloses a data access interface unit comprising: a physical storage device controller for receiving a first control signal from a first storage virtualization controller, and accordingly determining the first storage virtualization controller as the primary controller, and generating a first selection signal; a selector for receiving the first selection signal, and accordingly selecting data and signals from the first storage virtualization controller; and a clock generation circuit for providing a dedicated clock signal to the physical storage device, where when the physical storage device controller receives a re-set signal from a second storage virtualization controller, the physical storage device controller determines the second storage virtualization controller as the new primary controller, and accordingly generates a second selection signal so as to control the selector to select data and signals from the second storage virtualization controller.

Abstract: An energy storage system and a control method thereof are provided. The control method includes steps of: (a) providing an energy storage system including N energy storage modules; (b) obtaining a first sequence and a second sequence by sorting the N energy storage units based on the quantity of electricity in descending order and ascending order respectively; (c) determining a required power of the power grid according to a grid frequency; (d) when the required power is positive, controlling first X energy storage units in the first sequence to discharge for collectively providing an electrical energy, having same magnitude with the required power, to the power grid; and (e) when the required power is negative, controlling first Y energy storage units in the second sequence to collectively receive an electrical energy, having same magnitude with the required power, from the power grid for charging.

Abstract: Multiple chip module (MCM) structures are described. In an embodiment, a module includes a first and second components on the top side of a module substrate, a stiffener structure mounted on the top side of the module substrate, and a lid mounted on the stiffener structure and covering the first component and the second component. The stiffener is joined to the lid within a trench formed in a roof of the lid.

Abstract: A package structure includes a first semiconductor die, a second semiconductor die, an insulating encapsulant and a redistribution layer. The insulating encapsulant laterally surrounds the first semiconductor die and the second semiconductor die, wherein the insulating encapsulant includes a first portion sandwiched in between the first semiconductor die and the second semiconductor die, the first portion has a first recessed part adjacent to an edge of the first semiconductor die, and a second recessed part adjacent to an edge of the second semiconductor die. The redistribution layer is disposed on and electrically connected to the first semiconductor die and the second semiconductor die.

Abstract: The invention discloses a pure end-to-end deep reinforcement learning for training car racing game AI bot that uses only the velocity information extracted from screen for both training and testing phases without using any internal state from game environment, such as the car facing angle. The learned AI bot can play better than the average performance of human players. In addition, the reward function is designed to consist only the velocity value, and use Ape-X distributed training framework combined with a variant of Deep Q Network to solve the sparse training signal problem caused by the reward function of an original design. Moreover, limit learner rate method is designed that improves the training efficiency and training performance. The AI bot trained in this way can achieve performance beyond the average human level and reach a level close to professional players.

Abstract: Some implementations are directed to a wireless receiver. In some implementations, the wireless receiver may include a receiver body encompassing one or more antenna elements, a cover removably coupled to the receiver body, and a mounting bracket removably coupled to the receiver body. In some implementations, at least one of the one or more antenna elements, the cover, or the mounting bracket is movable with respect to the receiver body in order to align the wireless receiver with a signal path.

Abstract: A decision variable calculation method allowing a of reverse derivation requester to calculate decision variables based on a target result by pre-trained models provided by some participants in federated learning. The method includes the steps of providing the target result to each pre-trained model participating in this method and allowing them to reversely derive the input parameters, forming a loss function based on the difference between each pre-trained model and target result, integrating all input parameters into a total input parameter, and integrating all loss functions into a total loss function. An optimization problem is then constructed by the total input parameters and the total loss function. The solution to the optimization problem is the required decision variables.

Abstract: A method for calculating feasible process parameters of the present invention constructs an optimization model by treating a trained predictive model as a function of process parameters to be determined, wherein the objective function is to minimize the difference between the function value and the target result, subject to the conditions that the process parameters to be determined must satisfy, either individually or in relation to each other. Moreover, in order to solve the constrained optimization problem, the method with penalty function and barrier function can be used to convert the constrained optimization problem into an unconstrained optimization problem to make it more convenient to solve.

Abstract: A method for calculating feasible process parameters to achieve a given process result and the method comprises the following steps of: providing a trained prediction model, obtained by machine learning of a dataset by a machine learning method, wherein the dataset comprises a plurality of samples, each of the samples comprises a plurality of sample parameters, and the trained predictive model is configured to input a plurality of input parameters and generate a prediction result corresponding to the input parameters; setting an expected result as the prediction result of the trained predictive model and providing at least one confirmed input parameters of the input parameters; and comparing the expected result, the at least one confirmed input parameters, and the sample parameters of the samples in the dataset by a reverse derivation algorithm to determine at least one non-confirmed input parameters of the input parameters.

Abstract: The present invention provides a method for calculating decision variables. A dummy layer is added at an input layer of a trained neural network predictive model. The dummy layer includes a plurality of artificial neurons respectively connected to a corresponding input terminal of the input layer for the trained predictive model by a newly established link. The input value of each artificial neuron is set to 1, the bias value of the activation function is set to 0, and the output of the activation function is set to 1 when the input of the activation function is 1. The initial weight value of the newly established link is selected and set, and the weight values can be considered as decision variables, wherein the weight values can have ranges or other inter-conditional restrictions.

Abstract: A federated learning contribution calculation method comprises the following steps: a plurality of participants collaboratively developing a federated aggregation model by federated learning method according to their own local datasets; excluding the participation of at least one first participant in all participants, and then the remained participants collaboratively developing a contribution model by federated learning method; and, comparing the value of the first contribution model and the value of the federated model to obtain the contribution of the at least one first participant. The method of the present invention is capable of calculating the contribution(s) of single participant or multiple participants in the federated learning by few additional information and few additional calculations, so as to achieve the fair profit sharing according to the contributions.

Abstract: A method for calculating decision variables is configured to calculate the unconfirmed decision variables. First, the method provides a trained predictive mode obtained by machine learning through a machine learning method on a dataset. Next, the method transforms the objective function of the trained predictive model from a constrained objective function to an unconstrained objective function. The method then solves the optimization problem of the unconstrained objective function, wherein the optimizer, trained with the trained predictive model, calculates gradients to facilitate the solution process. Additionally, the samples from the dataset used to train the trained predictive model can be utilized to determine the initial samples for solving the optimization problem. The method for calculating decision variables of the invention can also add a dummy layer in front of the trained predictive model.

Abstract: A method and an electronic device of updating a neural network model are provided. The method includes the following steps. The neural network model is received. The neural network model includes a first neuron and a second neuron. The training data is input to the first neuron to output a first estimated value from the second neuron. A first weight of the first neuron is quantized to generate a third neuron, and a second weight of the second neuron is quantized to generate a fourth neuron. The training data is input to the third neuron to output a second estimated value from the fourth neuron. A first activation function of the first neuron and a second activation function of the second neuron are updated according to the first estimated value and the second estimated value, thereby generating the updated neural network model. The updated neural network model is output.

Abstract: Apparatus, systems, and methods for data movement in a memory device are described. In one embodiment, a memory controller comprises logic to move a row of data from a first row of a memory in a first section of a memory device to a second row of memory in a second section of the memory device without passing the data through a communication interface. Other embodiments are also disclosed and claimed.

Abstract: A keyboard module adapted to be used in an electronic device. The electronic device has a casing. The casing has a containing opening. The keyboard module contained in the containing opening includes a bottom plate, an elastic component, a cover and a key cap disposed on the elastic component. The bottom plate disposed within the casing has a circuit. The elastic component is disposed on the bottom plate or the circuit. The cover movably disposed in the containing opening is located above the bottom plate. The cover has an opening aligned to the elastic component. A portion of the key cap protrudes from the opening. Another portion of the key cap interferes with the opening, such that the key cap resists elasticity of the elastic component and moves to shorten a distance between the key cap and the bottom plate when the cover moves toward the bottom plate.

Abstract: A transceiver system having a phase and frequency locked architecture is described. The transceiver system includes a clock and data recovery type receiver, a frequency divider and a transmitter. The clock and data recovery type receiver receives an external signal from a host unit and extracts the external signal to generate a clock signal and a data signal. The frequency divider is used to divide the frequency of the clock signal for generating a reference clock signal. The transmitter transmits output data content based on the reference clock signal.

Abstract: A method and apparatus for performing clock extraction are provided. The method includes: performing edge analysis on a Training Sequence Equalization (TSEQ) pattern carried by a set of received signals that are received from a Universal Serial Bus (USB) port of an electronic device, to dynamically generate a plurality of analysis results; and performing frequency calibration on a frequency of an output clock of a Numerically Controlled Oscillator (NCO) according to a frequency that different types of analysis results within the plurality of analysis results alternatively occur, to utilize the output clock as a reference clock after completing the frequency calibration. More particularly, the method further includes: generating a set of de-multiplexed signals respectively corresponding to a plurality of bits, to perform the edge analysis by comparing respective voltage levels of de-multiplexed signals corresponding to every two adjacent bits of the plurality of bits within the set of de-multiplexed signals.

Abstract: Apparatus, systems, and methods for data movement in a memory device are described. In one embodiment, a memory controller comprises logic to move a row of data from a first row of a memory in a first section of a memory device to a second row of memory in a second section of the memory device without passing the data through a communication interface. Other embodiments are also disclosed and claimed.

Abstract: A keyboard module adapted to be used in an electronic device. The electronic device has a casing. The casing has a containing opening. The keyboard module contained in the containing opening includes a bottom plate, an elastic component, a cover and a key cap disposed on the elastic component. The bottom plate disposed within the casing has a circuit. The elastic component is disposed on the bottom plate or the circuit. The cover movably disposed in the containing opening is located above the bottom plate. The cover has an opening aligned to the elastic component. A portion of the key cap protrudes from the opening Another portion of the key cap interferes with the opening, such that the key cap resists elasticity of the elastic component and moves to shorten a distance between the key cap and the bottom plate when the cover moves toward the bottom plate.