Abstract: Decentralized bilevel optimization techniques for personalized learning over a heterogenous network are provided. In one aspect, a decentralized learning system includes: a distributed machine learning network with multiple nodes, and datasets associated with the nodes; and a bilevel learning structure at each of the nodes for optimizing one or more features from each of the datasets using a decentralized bilevel optimization solver, while maintaining distinct features from each of the datasets. A method for decentralized learning is also provided.

Abstract: Devices and methods are described for reducing etching due to Galvanic Effect within a printed circuit board (PCB) that may be used in an electronic device. Specifically, a contact trace is coupled to a contact finger that has a substantially larger surface area than the contact trace. The contact finger is configured to couple the electronic device to a host device. The contact trace is electrically isolated from the rest of the PCB circuitry during a fabrication process by a separation distance between an exposed portion of the contact trace and an impedance trace. The contact finger and the exposed portion of the contact trace are plated with a common material to reduce galvanic etching of the contact trace during fabrication. The contact trace is then connected to the impedance trace using a solder joint.

Abstract: The present disclosure discloses a stable and high-capacity neutral aqueous redox flow lithium battery based on redox-targeting reaction and belongs to the technical field of flow lithium batteries. The present disclosure solves the technical problem that an existing flow battery can only work at low current density. The flow lithium battery of the present disclosure includes a positive electrode storage tank and a negative electrode storage tank; the positive electrode storage tank is filled with a positive electrolyte; and the negative electrode storage tank is filled with a negative electrolyte. The flow lithium battery is characterized in that the positive electrolyte includes a salt containing [Fe(CN)6]4? and/or [Fe(CN)6]3?, and the positive electrode storage tank is further filled with LFP particles and/or FP particles. The flow lithium battery of the present disclosure has wide application prospects in the field of large-scale energy storage.

Abstract: A semiconductor device includes a substrate, semiconductor dies on the substrate, molding compound and reinforcing blocks on the substrate. The reinforcing blocks may be provided at positions on the substrate where mechanical stresses develop in the device during singulation, such as at curves and/or discontinuous points around the outline of the substrate, to add strength to the substrate.

Abstract: In one example, the present application describes a Printed Circuit Board (PCB) that mitigates galvanic corrosion during an Organic Solderability Preservative (OSP) process used during fabrication of the PCB. The PCB includes a first metal pattern and a second metal pattern electrically coupled to each other, where the first and second metal patterns are different metals. The first metal pattern has a first area that is exposed by a solder mask layer, and the second metal pattern has a second area that is exposed by the solder mask area. A ratio of the first area to the second area is less than a threshold ratio to mitigate the galvanic corrosion of the second metal pattern exposed on the PCB during the OSP process.

Abstract: Devices and methods are described for reducing etching due to Galvanic Effect within a printed circuit board (PCB) that may be used in an electronic device. Specifically, a contact trace is coupled to a contact finger that has a substantially larger surface area than the contact trace. The contact finger is configured to couple the electronic device to a host device. The contact trace is electrically isolated from the rest of the PCB circuitry during a fabrication process by a separation distance between an exposed portion of the contact trace and an impedance trace. The contact finger and the exposed portion of the contact trace are plated with a common material to reduce galvanic etching of the contact trace during fabrication. The contact trace is then connected to the impedance trace using a solder joint.

Abstract: Methods and systems for training a model and automated motion include learning Markov decision processes using reinforcement learning in respective training environments. Logic rules are extracted from the Markov decision processes. T reward logic neural network (LNN) and a safety LNN are trained using the logic rules extracted from the Markov decision processes. The reward LNN and the safety LNN each take a state-action pair as an input and output a corresponding score for the state-action pair.

Abstract: A reinforcement learning system includes a plurality of agents, each agent having an individual reward function and one or more safety constraints that involve joint actions of the agents, wherein each agent maximizes a team-average long-term return in performing the joint actions, subject to the safety constraints, and participates in operating a physical system. A peer-to-peer communication network is configured to connect the plurality of agents. A distributed constrained Markov decision process (D-CMDP) model is implemented over the peer-to-peer communication network and is configured to perform policy optimization using a decentralized policy gradient (PG) method, wherein the participation of each agent in operating the physical system is based on the D-CMDP model.

Abstract: The present disclosure discloses a stable and high-capacity neutral aqueous redox flow lithium battery based on redox-targeting reaction and belongs to the technical field of flow lithium batteries. The present disclosure solves the technical problem that an existing flow battery can only work at low current density. The flow lithium battery of the present disclosure includes a positive electrode storage tank and a negative electrode storage tank; the positive electrode storage tank is filled with a positive electrolyte; and the negative electrode storage tank is filled with a negative electrolyte. The flow lithium battery is characterized in that the positive electrolyte includes a salt containing [Fe(CN)6]4? and/or [Fe(CN)6]3?, and the positive electrode storage tank is further filled with LFP particles and/or FP particles. The flow lithium battery of the present disclosure has wide application prospects in the field of large-scale energy storage.

Abstract: Scalable distributed adversarial training techniques for robust deep neural networks are provided. In one aspect, a method for adversarial training of a deep neural network-based model by distributed computing machines M includes, by distributed computing machines M: obtaining adversarial perturbation-modified training examples for samples in a local dataset D(i); computing gradients of a local cost function fi with respect to parameters ? of the deep neural network-based model using the adversarial perturbation-modified training examples; transmitting the gradients of the local cost function fi to a server which aggregates the gradients of the local cost function fi and transmits an aggregated gradient to the distributed computing machines M; and updating the parameters ? of the deep neural network-based model stored at each of the distributed computing machines M based on the aggregated gradient received from the server.

Abstract: A trained machine learning model and a training dataset used to train the trained machine learning model can be received. Based on the training dataset, unsupervised adversarial examples can be generated. Robustness of the trained machine learning model can be determined using the generated unsupervised adversarial examples. The training dataset can be augmented with the generated unsupervised adversarial examples. The trained machine learning model can be retrained using the augmented training dataset.

Abstract: Devices and methods are described for reducing etching due to galvanic effect within a printed circuit board that may be used, for example, in a data storage device, such as a card-type data storage device. Specifically, a contact trace is coupled to a contact finger that has a substantially larger surface area than the contact trance, and that is configured to couple the data storage device to a host device. The contact trace is electrically isolated from the rest of the circuitry during a fabrication process. The contact finger and an exposed portion of the contact trace are plated with a common material to reduce galvanic etching of the contact trace during fabrication. The contact trace is then connected to an impedance trace though at least one of a component and a bond wire.

Abstract: A substrate is disclosed having a stress relief layer. The stress relief layer may be applied to a dielectric core of the substrate, beneath a conductive layer in which electrical traces and contact pads are formed. The substrate including the stress relief layer may be incorporated into a semiconductor product which may, for example, be mounted on a host printed circuit board using solder balls on a surface of the substrate. The stress relief layer helps dissipate stresses within the substrate and improves the board level reliability.

Abstract: A cooling appliance includes a cabinet, first and second doors connected to the cabinet and a door-opening detection device having a magnetic element disposed on the first door and a magnetic sensitive element disposed on the cabinet. A first magnetic conductive element on the second door can be magnetized by the magnetic element. The magnetic sensitive element can produce a door-opening signal based on a magnetic field of the first magnetic conductive element. The magnetic sensitive element produces the door-opening signal when at least one of the doors is open. In this way it can be detected, by using one suite of door-opening detection devices, if any door among multiple doors is in an open state. The structure is simple, and the number of components can be reduced. Moreover, wires are reduced, which lowers manufacturing costs and difficulty.

Abstract: A door for a refrigerator includes a door housing, which is provided with a thermal insulation material therein, where the thermal insulation material is combined with the door housing. The door housing has a concave portion recessed backwards from a front surface of the door, where one end of the concave portion is provided with a ramp that extends forwards. A detachable member is provided and is isolated from the thermal insulation material and is connected to the door housing. Accordingly, at least one part of the ramp is formed by the detachable member.

Abstract: A door for a refrigerator includes a door housing, which is provided with a thermal insulation material therein, where the thermal insulation material is combined with the door housing. The door housing has a concave portion recessed backwards from a front surface of the door, where one end of the concave portion is provided with a ramp that extends forwards. A detachable member is provided and is isolated from the thermal insulation material and is connected to the door housing. Accordingly, at least one part of the ramp is formed by the detachable member.

Abstract: A cooling appliance includes a cabinet, first and second doors connected to the cabinet and a door-opening detection device having a magnetic element disposed on the first door and a magnetic sensitive element disposed on the cabinet. A first magnetic conductive element on the second door can be magnetized by the magnetic element. The magnetic sensitive element can produce a door-opening signal based on a magnetic field of the first magnetic conductive element. The magnetic sensitive element produces the door-opening signal when at least one of the doors is open. In this way it can be detected, by using one suite of door-opening detection devices, if any door among multiple doors is in an open state. The structure is simple, and the number of components can be reduced. Moreover, wires are reduced, which lowers manufacturing costs and difficulty.

Abstract: A light intensity adjustable ultraviolet device for curing an optical fiber coating includes a cylindrical mounting base; a UVLED light source module mounted along a peripheral direction and an axial direction in an inner cavity of the cylindrical mounting base; a cylindrical focusing lens configured in front of a light emitting surface of the UVLED light source module, so that ultraviolet light emitted by the UVLED light source module is focused on a curing axis; and an ultraviolet sensor mounted in the inner cavity of the cylindrical mounting base, wherein the ultraviolet sensor is connected to a UVLED power supply control module via an ultraviolet intensity signal processing module; the UVLED power supply control module is connected to the UVLED light source module, so that an optical fiber drawing speed and an ultraviolet intensity form a control closed loop.

Abstract: A light intensity adjustable ultraviolet device for curing an optical fiber coating includes a cylindrical mounting base; a UVLED light source module mounted along a peripheral direction and an axial direction in an inner cavity of the cylindrical mounting base; a cylindrical focusing lens configured in front of a light emitting surface of the UVLED light source module, so that ultraviolet light emitted by the UVLED light source module is focused on a curing axis; and an ultraviolet sensor mounted in the inner cavity of the cylindrical mounting base, wherein the ultraviolet sensor is connected to a UVLED power supply control module via an ultraviolet intensity signal processing module; the UVLED power supply control module is connected to the UVLED light source module, so that an optical fiber drawing speed and an ultraviolet intensity form a control closed loop.

Abstract: An ice dispenser for a refrigerator is provided. The ice dispenser includes an ice storage container housing and an ice generator having a hollow mold which is filled with water. The ice storage container housing and the ice generator are combined into one modular unit.