Abstract: A multidimensional knowledge transfer model for predicting CTR of a computational ad, the multidimensional knowledge transfer model comprises: a pre-processor for building an ad group node graph based on similarities among ad group nodes, an ad campaign node graph from merging the ad group node graph, and an ad account node graph from merging the ad campaign node graph. The multidimensional knowledge transfer model further comprises a multi-knowledge CTR prediction model for each of the ad account, ad campaign, and ad group layers. The multi-knowledge CTR prediction model predicts the respective node's CTR from the ad account node graph, ad campaign node graph, or ad group node graph, features of the audience group, and features of the node having its CTR predicted appended with the hidden vector of its parent node extracted from the upper layer multi-knowledge CTR prediction model.

Abstract: A power delivery architecture is described that improves system voltage conversion and operational efficiency. The power delivery architecture performs monitoring of various system conditions such as a current power state, a current power policy setting, workload conditions, component temperatures, a state of charge of a battery, etc. The power delivery architecture may adjust the power profile provided by a power delivery source, which may result in an adjustment to the VBUS voltage and/or current when a USB-based power delivery architecture is implemented. The power delivery architecture may also adjust a mode of operation of an onboard battery charger.

Abstract: The invention relates to a heat management structure with graphene and copper, and a formation method thereof, comprising a copper foil layer provided, then forming a graphene layer on the copper foil layer surface, and forming an electroplating copper layer on the graphene layer surface, and eventually forming an electroplating copper layer-graphene layer-copper foil layer sandwich heat management structure.

Abstract: Random token segmentation may be implemented for next token prediction. Text data may be received for training a machine learning model to predict a next token given input text tokens. Multiple tokens may be determined from the text data. Different ones of the multiple token may be randomly segmented in to sub-tokens. The machine learning model may then be trained using the multiple tokens including the respective sub-tokens as a training data set.

Abstract: Embodiments herein relate to avoiding damage to a transistor in a power-sinking device that receives power from an external power source via a Universal Serial Bus port. In one aspect, a controller of the device sets a current limit to a reduced level during a wait period after the external power source is connected to the power-sinking device. The wait period avoids damage to the transistor by allowing its input and output voltages to equalize before the current is increased. Upon expiration of the wait period, the current limit is increased to a level negotiated with the external power source. Other aspects involve considering a sleep or low/dead battery state of the power-sinking device. The current limit can be set by programming a current limit of a battery charger coupled to between the transistor and a power bus of the device.

Abstract: Embodiments herein relate to a power monitor that can be used to dynamically change a power level of an electronic device and/or operational settings of a processor of the electronic device. Specifically, the power monitor may be configured to identify “droop” of power, and logic to update the power level and/or operational settings in accordance with identification of the droop. Other embodiments may be described and claimed.

Abstract: A single layer process is utilized to reduce swing effect interference and reflection during imaging of a photoresist. An anti-reflective additive is added to a photoresist, wherein the anti-reflective additive has a dye portion and a reactive portion. Upon dispensing the reactive portion will react with underlying structures to form an anti-reflective coating between the underlying structure and a remainder of the photoresist. During imaging, the anti-reflective coating will either absorb the energy, preventing it from being reflected, or else modify the optical path of reflection, thereby helping to reduce interference caused by the reflected energy.

Abstract: A photoresist apparatus and a method are provided. The photoresist apparatus includes a pre-baking apparatus. The pre-baking apparatus includes: a hot-plate, a first cover over the hot-plate, a second cover over the first cover, a first heating element extending along a topmost surface of the first cover, and a second heating element extending along a topmost surface of the second cover.

Abstract: Disclosed herein are vaccine compositions, in particular, polyvalent icosahedral compositions for antigen presentation. The disclosed compositions may contain an S particle made up of recombinant fusion proteins. The recombinant fusion proteins may include a norovirus (NoV) S domain protein, a linker protein domain operatively connected to the norovirus S domain protein, and an antigen protein domain operatively connected to said linker.

Abstract: A package includes a device die, an encapsulant encapsulating the device die therein, a first plurality of through-vias penetrating through the encapsulant, a second plurality of through-vias penetrating through the encapsulant, and redistribution lines over and electrically coupling to the first plurality of through-vias. The first plurality of through-vias include an array. The second plurality of through-vias are outside of the first array, and the second plurality of through-vias are larger than the first plurality of through-vias.

Abstract: A package includes a device die, an encapsulant encapsulating the device die therein, a first plurality of through-vias penetrating through the encapsulant, a second plurality of through-vias penetrating through the encapsulant, and redistribution lines over and electrically coupling to the first plurality of through-vias. The first plurality of through-vias include an array. The second plurality of through-vias are outside of the first array, and the second plurality of through-vias are larger than the first plurality of through-vias.

Abstract: Systems, apparatuses and methods may provide for technology that allocates a portion of operational power in a source device to an external sink device in response to a connection of the external sink device to the source device, detects a low power state with respect to the external sink device, and decreases the portion of operational power allocated to the external sink device in response to the low power state.

Abstract: The present disclosure discloses an air passage unit for a massaging pool. The air passage unit comprises an air passing portion, a first air connecting portion, and a first fastening structure. The first air connecting portion comprises two first air connecting openings configured to be in communication with the air passing portion, the first fastening structure is configured to be detachably connected to the massaging pool, and the two first air connecting openings are disposed on the first fastening structure so as to enable the two first air connecting openings to be synchronously connected to or detached from the massaging pool through the first fastening structure being connected to or detached from the massaging pool. The present disclosure further discloses an air passage system for a massaging pool. The air passage system comprises an air pump, and an air passage unit. The air passage unit comprises an air passing portion, a first air connecting portion, and a control mechanism.

Abstract: Embodiments herein relate to an electronic circuit with a first port for providing a first power signal and a second port for providing a second power signal. Each of the ports may be coupled with a respective system gate and charging gate. A system gate may selectively couple a port with a system load, while the charging gate may selectively couple a port with a battery. Other embodiments may be described and/or claimed.

Abstract: Embodiments herein relate to a computer which operates in a normal power mode, a full low power mode and a modified low power mode. A power supply unit (PSU) includes main power rails and a standby power rail. An intelligent decision is made as to when to turn on and off the main power rails based on the needs of connected devices such as USB Type-C devices. Power Delivery controllers communicate with the USB devices to determine their power consumption needs, and the total power consumption is aggregated at an embedded controller. If the total exceeds an available power budget, the PSU is controlled to allow for a modified low power mode in which full power is maintained at the main power rails while non-essential internal components are turned off.

Abstract: Systems and techniques that facilitate few-shot temporal action localization based on graph convolutional networks are provided. In one or more embodiments, a graph component can generate a graph that models a support set of temporal action classifications. Nodes of the graph can correspond to respective temporal action classifications in the support set. Edges of the graph can correspond to similarities between the respective temporal action classifications. In various embodiments, a convolution component can perform a convolution on the graph, such that the nodes of the graph output respective matching scores indicating levels of match between the respective temporal action classifications and an action to be classified. In various embodiments, an instantiation component can input into the nodes respective input vectors based on a proposed feature vector representing the action to be classified.

Abstract: An above ground frame pool (10) includes a frame assembly (20) and a liner (40) supported by the frame assembly (20). The frame assembly (20) includes a plurality of horizontal pipes (22), a plurality of vertical pipes (24), and a plurality of T-shaped joints (30). The joints (30) may lack mechanical fasteners with adjacent pipes or corresponding apertures. The joints (30) may be one-piece or multi-piece structures. Such joints (30) maybe water-resistant and convenient to manufacture, assemble, disassemble, and store.

Abstract: A deep learning module classifies messages received from a plurality of entities into one or more conversation threads. In response to receiving a subsequent message, the deep learning module determines which of the one or more conversation threads and a new conversation thread is contextually a best fit for the subsequent message. The subsequent message is added to the determined conversation thread.

Abstract: The invention is to provide a protein measurement apparatus and operation method thereof, including a host device, a display device, and a sample preparation device, wherein the host device includes a power supply, an electromagnetic-wave generator, an electromagnetic-wave detection circuit, a micro-controller and a display device. The operation method of the protein measurement apparatus includes starting the micro-electromagnetic-wave generator to generate an electromagnetic-wave signal and transmit the electromagnetic-wave signal to the sample preparation device; collecting the electromagnetic-wave reflection signal, which is reflected from the sample preparation device; detecting and analyzing the electromagnetic-wave reflection signal by using the electromagnetic-wave detection circuit; and comparing and analyzing the above-mentioned electromagnetic-wave reflection signal with characteristic curve by the microcontroller, so that the insulin concentration of a tester's saliva can be obtained.