Abstract: The systems and techniques described herein relate to predicting user conversions in online advertising. Input data associated with user and advertisement features may be processed through neural networks to generate embedding representations or feature cross representations. A multi-task layer calculates probabilities associated with multiple user actions like clicks, page views, sign-ups, or purchases. Click-through and view-through conversion probabilities may be calculated to generate a score. The systems and techniques described herein perform predictions on multiple types of user actions despite data sparsity and negative transfer challenges, enhancing advertisement targeting and improving conversion metrics.

Abstract: A display device and a display method are provided. The display device includes: a driving circuit; a first control circuit configured to, in response to a selection signal from the driving circuit being at a first level, send a first data signal to the driving circuit via the first data line so that the driving circuit drives the display panel to display in a first display mode; and a second control circuit configured to, in response to the selection signal being at a second level, send a second data signal to the driving circuit via the second data line so that the driving circuit drives the display panel to display in a second display mode. A display power of the display panel in the first display mode is greater than that in the second display mode.

Abstract: Disclosed in the present invention are a series of tricyclic compounds, and particularly disclosed are a compound shown in formula (I) or a pharmaceutically acceptable salt thereof.

Abstract: The present application provides a self-supervised learning method performed by a computer device. The method includes: performing a data enhancement on an original medical image to obtain a first enhanced image and a second enhanced image, the first enhanced image and the second enhanced image being positive samples of each other; performing feature extractions on the first enhanced image and the second enhanced image by a feature extraction model to obtain a first image feature of the first enhanced image and a second image feature of the second enhanced image; determining a model loss of the feature extraction model based on the first image feature, the second image feature, and a negative sample image feature, the negative sample image feature being an image feature corresponding to other original medical images; and training the feature extraction model based on the model loss.

Abstract: The present invention provides an application of ?-asarone in the preparation of a medicine for preventing or treating hemorrhagic strokes. ?-asarone has the structure shown in formula I, and may significantly improve short-term neurological deficits and long-term learning and memory functions of model rats, reduce cerebral edema, improve the permeability of the blood-brain barrier, and prevent or mitigate the atrophy of brain tissue during a recovery period. ?-asarone has a precise therapeutic effect on animal models of hemorrhagic stroke without obvious toxic side effects, and is expected to have great application prospects as a medicine for preventing/treating hemorrhagic strokes.

Abstract: A semiconductor device package comprises a semiconductor device, a first encapsulant surrounding the semiconductor device, a second encapsulant covering the semiconductor device and the first encapsulant, and a redistribution layer extending through the second encapsulant and electrically connected to the semiconductor device.

Abstract: Systems and methods for selecting items from a pool of items based on comparisons of composite embeddings are disclosed. A first initial embedding is obtained from a first database for each item in a pool of items and a second initial embedding is obtained from a second database for each item in the pool of items. The first initial embedding is generated using a first embedding model and the second initial embedding is generated using a second embedding model. A first composite embedding is generated for each item in the pool of items comprising the first initial embedding and the second initial embedding; and compare the first composite embedding for each item in the pool of items to a first anchor embedding, wherein the first anchor embedding comprises a first initial anchor embedding generated using the first embedding model and a second initial anchor embedding generated using the second embedding model.

Abstract: The present disclosure provides an extended screen control method, and a computer-readable storage medium. The method comprises: obtaining control information corresponding to the extended screen, according to a motion type of the terminal device, in which the motion type of the terminal device is determined according to motion data collected by the terminal device; and controlling the extended screen according to the control information.

Abstract: Examples described herein provide a computer-implemented method for reducing storage replication by operation logs in cloud multiple zone environments. The method includes deploying, at an operator engine, a custom resource, the custom resource defining multiple roles. The method further includes parsing the custom resource and calculating a calculated checksum for each of the multiple roles. The method further includes, for each of the multiple roles, comparing the calculated checksum to an expected checksum to determine whether a change in checksum has occurred. The method further includes, responsive to determining that the change in checksum has occurred for at least one of the roles of the multiple roles, executing role tasks for each of the at least one of the roles of the multiple roles for which the change in checksum has occurred.

Abstract: A motor control unit, a powertrain, and an anti-slip control method for an electric vehicle are described. The motor control unit is configured to control one drive motor to drive an electric vehicle and control a plurality of wheel end brake apparatuses of the electric vehicle, and the plurality of wheel end brake apparatuses are respectively configured to brake a plurality of wheels of the electric vehicle. The motor control unit is further configured to: in a process of controlling the one drive motor to drive the electric vehicle, in response to slipping of one wheel in the plurality of wheels, control one wheel end brake apparatus corresponding to the one wheel to brake the one wheel. Based on this, when one wheel of the electric vehicle slips, the motor control unit can perform control, so as to suppress the slipping of the wheel more quickly.

Abstract: Systems and methods for selecting items from a pool of items based on comparisons of composite embeddings are disclosed. An anchor embedding is generated for a target string. The anchor embedding is a composite embedding including at least a first initial embedding and a second initial embedding. An item embedding is obtained for each item in a pool of items. Each item embedding is a composite embedding including a first initial item embedding and a second initial item embedding. A similarity score is generated by comparing the item embedding to the anchor embedding for each item in the pool of items and a set of items is selected from the pool of items. The set of items includes a predetermined number of items in the pool of items having a highest similarity score.

Abstract: A method for generating a motion trajectory of a device is provided. The method includes: obtaining a first coordinate of a wearable device, and obtaining step data and direction data generated by the wearable device; determining a second coordinate of the wearable device based on the step data and the direction data; and generating a motion trajectory of the wearable device based on the first coordinate and the second coordinate.

Abstract: The systems and techniques described herein relate to predicting user conversions in online advertising. Input data associated with user and advertisement features may be processed through neural networks to generate embedding representations or feature cross representations. A multi-task layer calculates probabilities associated with multiple user actions like clicks, page views, sign-ups, or purchases. Click-through and view-through conversion probabilities may be calculated to generate a score. The systems and techniques described herein perform predictions on multiple types of user actions despite data sparsity and negative transfer challenges, enhancing advertisement targeting and improving conversion metrics.

Abstract: Described is a system for knowledge distillation in ad machine learning models by training a plurality of machine learning models on respective datasets collected over respective time periods; collecting a first dataset comprising ad impression data and ad conversion data over a first time period; applying each of the first dataset to the plurality of machine learning models to generate a plurality of labels; derive a value based on the plurality of labels; training a first machine learning model based on the application of the first dataset to the first machine learning model; applying a plurality of ads to the trained first machine learning model to receive individual predicted ad conversion rates for each of the plurality of ads; and ranking the ads based on the predicted ad conversion rates.

Abstract: Described is a system for training a machine learning model by collecting a first dataset comprising ad impression data and ad conversion data over a first time period; training a first machine learning model using the first dataset; collecting a second dataset comprising ad impression data and ad conversion data over a second time period; selecting a subset of the first dataset; training a second machine learning model using a combined dataset of the subset of the first dataset and the second dataset to generate a second trained machine learning model configured to generate predicted ad conversion rates for new ads; applying a plurality of ads to the second trained machine learning model to receive individual predicted ad conversion rates for each of the plurality of ads; and ranking the ads based on the predicted ad conversion rates.

Abstract: The invention discloses a compound represented by the following general formula (I) or a pharmaceutically acceptable salt thereof. The invention also discloses a preparation method of the compound and use thereof in preventing and/or treating cancer or tumor-related diseases, in particular diseases such as breast cancer, prostate cancer, lung cancer, multiple myeloma, leukemia, gastric cancer, ovarian cancer, colon cancer, liver cancer, pancreatic cancer, human glioma and the like. The compound of the invention is expected to be developed into a new generation of anticancer drugs.

Abstract: Provided are a battery pack casing and a battery pack. The battery pack casing includes an enclosure, an upper cover plate, a bottom protective plate, and at least one pressure relief valve. The bottom protective plate includes an upper plate, a lower plate, and multiple connecting platforms to form a pressure relief chamber. The upper plate also has multiple pressure relief holes that communicate with the pressure relief chamber. A buffer chamber that communicates with the pressure relief chamber is disposed in the enclosure. The at least one pressure relief valve is disposed on the enclosure and communicates with the buffer chamber.

Abstract: A semiconductor device package comprises a semiconductor device, a first encapsulant surrounding the semiconductor device, a second encapsulant covering the semiconductor device and the first encapsulant, and a redistribution layer extending through the second encapsulant and electrically connected to the semiconductor device.

Abstract: The present disclosure discloses a normal load testing device for a pump body, which includes a fixed platform. A rotating mechanism is provided on the fixed platform, a rotating end of the rotating mechanism is connected to a fixed plate. The normal load testing device further includes: a plurality of connecting heating rings configured to connect the pump body to the fixed plate and heat the pump body; a stabilizing mechanism, configured to strengthen the stability of the pump body and the fixed plate after connection; a sensor, configured to collect test data of the pump body during the test; and a hammering mechanism, configured to excite the pump body from top to bottom. The normal load testing device can simulate the connection form between the pump body and an end cover, and can test a plurality of pump bodies at the same time, thus saving the test time.

Abstract: A display module, a display device, and a control method for the display module. The display module comprises a driving module (1) and a display panel (2). The driving module (1) comprises a data signal output unit (11). The display panel (2) comprises N rows of sub-pixels. A scanning stage of each row of sub-pixels comprises a data writing period and a reserved period immediately following the data writing period; and the data signal output unit (11) is configured to be in an off state in the data writing period of the n-th row of sub-pixels (610), and switch from the off state to an on state in the reserved period of the n-th row of sub-pixels, so as to provide a stable first data signal for the (n+1)-th row of sub-pixels in the data writing period of the (n+1)-th row of sub-pixels, wherein 0<n<N (620).