Abstract: Disclosed is a method for manufacturing anti-glare glass, including the following steps: performing surface cleaning on a glass substrate; depositing a film layer on a surface of the glass substrate; coating the film layer with photoresist; carrying out exposure, development and film layer etching; performing dry etching to directionally etch a longitudinal depth of the glass, followed by wet etching to simultaneously etch in both longitudinal and transverse directions to obtain a specific aspect ratio and increase an anti-glare area; or performing wet etching to construct a preliminary morphology on the glass surface, followed by dry etching to achieve precise depth control and feature shaping in specific areas. The method can effectively improve anti-glare performance and approach the disordered diffusion effect of frosted glass; and the surface texture distribution is more uniform, overcoming the defects of the conventional frosting process.

Abstract: A training apparatus for a machine learning model includes at least one processor configured to: acquire a plurality of two-dimensional images and a machine learning model to be trained, where the plurality of two-dimensional images include a target, and the machine learning model includes first and second three-dimensional pose estimation modules; based on any one of the two-dimensional images, use the first three-dimensional pose estimation module to estimate a first three-dimensional pose of the target; based on at least two of the two-dimensional images, use the second three-dimensional pose estimation module to estimate a second three-dimensional pose of the target, where the at least two of the two-dimensional images have different viewing angles from each other; and train, based on a difference between the first and second three-dimensional poses, at least one of the first three-dimensional pose estimation module or the second three-dimensional pose estimation module.

Abstract: Methods, systems, and computer program products are provided for multi-layer analysis and detection of vulnerability of machine learning models to adversarial attacks. An example method includes selecting a sample, inputting the sample to a machine learning model, perturbing the sample to generate perturbed samples, inputting each respective perturbed sample to the machine learning model, and determining a smallest perturbation distance for the input layer. For each respective hidden layer of the machine learning model, a respective hidden layer representation of the sample is perturbed to generate perturbed hidden layer representations. Each respective perturbed hidden layer representation is inputted to the machine learning model, and a respective smallest perturbation distance for each respective hidden layer is determined. At least one vulnerability metric is determined based on the smallest perturbation distance for the input layer and the hidden layer(s).

Abstract: False data entities attempt to evade getting caught by changing their information repeatedly over time. Systems and methods are provided to detect false data entities. A computing system ingests a plurality of data files respectively from a plurality of external data sources using the network interface within a current specified time period. It consolidates a plurality of data entries of false data entities from the plurality of data files into a consolidated list for the current specified time period. It compares the consolidated list for the current specified time period with a previous consolidated list corresponding to a previous specified time period to identify one or more differences. It then displays, using a graphical user interface (GUI), the one or more differences in association with a given false entity.

Abstract: The present application provides a display panel and a display device. A light-emitting device layer includes an n-type charge generation layer and a p-type charge generation layer. The light-emitting device layer has a first operating voltage to operate at a preset current density, and has a second operating voltage after operating at the preset current density and a first preset temperature for a preset time. The absolute value of a difference value between the first operating voltage and the second operating voltage is a1, a1 is less than or equal to 1V. The first preset temperature is greater than or equal to 50° C.

Abstract: A method, system, and computer program product for managing model updates at multiple data centers hosting a same machine learning model obtain a plurality of first feature profiles input to a first implementation of a first machine learning model and a plurality of first model states determined from processing a model input with the first implementation; determine that a first model policy associated with the first machine learning model is satisfied, based on the plurality of first model states including a set of model states defined by the first model policy and the plurality of first feature profiles including a set of feature profiles defined by the first model policy; and in response to determining that the first model policy is satisfied, provide the plurality of first model states and the plurality of first feature profiles for updating at least one second implementation of the first machine learning model.

Abstract: The present disclosure provides a data sending method and apparatus. The method may include: determining one or more second sending policies of one or more users at a target transmission time interval (TTI) based on data amount and one or more first sending policies of the one or more users, so that a second sending policy of a first user is different from a first sending policy of the first user. In addition, second energy consumption corresponding to the one or more second sending policies is less than first energy consumption corresponding to the one or more first sending policies, an occupied second time-frequency resource under the one or more second sending policies is greater than an occupied first time-frequency resource under the one or more first sending policies, and the second time-frequency resource is less than an available time-frequency resource provided by a base station at the target TTI.

Abstract: A display panel and a display device are provided. An organic light-emitting layer includes a doped region and an undoped region, and the undoped region is located between the doped region and the hole blocking layer. When the display panel is in a low grayscale display state, an exciton concentration distribution curve of the organic light-emitting layer includes a first peak located in the doped region and a second peak located in the undoped region, and a peak value of the second peak is less than a peak value of the first peak.

Abstract: Provided is a system for testing a machine learning (ML) model based on simulations in an offline environment that includes at least one processor programmed or configured to receive historical transaction data, generate online simulation data, wherein generating the online simulation data includes modifying the historical timestamp of each data record to provide online simulation data. The processor is further programmed or configured to determine a timeline for a plurality of data insertion actions and a plurality of data request actions based on the online simulation data, perform a simulation of online activities involving a stateful ML model using the timeline for the plurality of data insertion actions and the plurality of data request actions, and validate the stateful ML model based on the simulation of online activities. Methods and computer program products are also provided.

Abstract: Disclosed in the present application are a microinverter, a photovoltaic system, and a control method. The microinverter comprises a controller, a primary-side H bridge, a transformer, a bidirectional switch bridge arm, and a capacitor bridge arm, wherein an input end of the primary-side H bridge is connected to a direct-current power supply; two output ends of the primary-side H bridge are respectively connected to two ends of a primary side of the transformer, and two ends of a secondary side of the transformer are respectively connected to a midpoint of the bidirectional switch bridge arm and a midpoint of the capacitor bridge arm; the bidirectional switch bridge arm and the capacitor bridge arm are connected in parallel; and the controller is used for switching control modes for the primary-side H bridge and the bidirectional switch bridge arm when the phase of each cycle of a power grid voltage is an odd multiple of 0.25?.

Abstract: Provided in the present application are a converter and a zero crossing point control method therefor. The zero crossing point control method comprises: first, determining whether the current power grid voltage is within a preset range of a zero crossing point; if the current power grid voltage is not within the preset range of the zero crossing point, controlling a primary-side circuit and a secondary-side circuit, which are connected to a transformer in a converter, to operate in a normal working mode; and if the current power grid voltage is within the preset range of the zero crossing point, controlling the primary-side circuit and the secondary-side circuit to operate in a preset working mode.

Abstract: Provided is a system that includes a processor to provide a first input to an autoencoder machine learning model; generate a first output of the autoencoder machine learning model based on the first input; provide the first input to a production machine learning model; provide the first output of the autoencoder machine learning model as a second input to the production machine learning model; generate a first output of the production machine learning model based on the first input; generate a second output of the production machine learning model based on the second input; determine a metric of divergence between the first output and the second output of the production machine learning model, wherein the metric of divergence comprises an indication of whether the first input is associated with an adversarial attack; and perform an action. Methods and computer program products are also provided.

Abstract: This application discloses material with dual heterogeneous structure, preparation method and application thereof. The material with dual heterogeneous structure is CdS/Bi2S3/ZnS composite material. With spherical Bi2S3 as the carrier, CdS particles and ZnS particles are uniformly dispersed on the surface of Bi2S3 to form a dual heterogeneous structure. The preparation method of the material with dual heterogeneous structure, comprising: dissolving Bi (NO3)3·5H2O in ethylene glycol solution to obtain Solution A; dissolving ZnCl2 in water to obtain solution B; dissolving CdCl2·2.5H2O in water to obtain solution C; mixing the solution A, B and C, and adding L-cysteine to the mixture. After hydrothermal reaction, washing and drying, CdS/Bi2S3/ZnS composite material is obtained. The material with dual heterogeneous structure can be applied in photocatalytic reduction of carbon dioxide, which selectively reduces carbon dioxide to CH4.

Abstract: Provided is a system for generating an inference based on real-time selection of a machine learning model using a machine learning model framework that includes at least one processor programmed or configured to receive a request for inference, wherein the request includes a payload, select a machine learning model of a plurality of machine learning models based on the request for inference, determine an aggregation of data based on the machine learning model and the payload of the request, transform the aggregation of data into inference data, wherein the inference data has a configuration that is capable of being processed by the machine learning model, and generate an inference based on the inference data using the machine learning model. Methods and computer program products are also provided.

Abstract: This application discloses Bi2S3/ZnS composites comprising S—Bi double defect and its preparation method. The preparation method includes: bismuth nitrate pentahydrate is dissolved in mannitol solution to obtain solution A; zinc chloride is dissolved in water to obtain solution B; solution A is mixed with solution B, L-cysteine is added, and then obtained by hydrothermal reaction. The method described in this application can obtain Bi2S3/ZnS composites containing both cationic Bi defects and anionic S defects. The composite material can inhibit the recombination of photogenerated electron-hole pairs while promoting the photogenerated charge transfer rate, and effectively improve the ability of photocatalytic reduction of carbon dioxide.

Abstract: The present disclosure provides a display panel and an electronic device. A difference value between a size of a first optical part and a size of a first sub-pixel of the display panel is defined as a first difference value. A difference value between a size of a second optical part and a size of a second sub-pixel of the display panel is defined as a second difference value. The first difference value corresponding to a maximum light-emitting efficiency of the first sub-pixel is defined as a first preset value. The second difference corresponding to a maximum light-emitting efficiency of the second sub-pixel is defined as a second preset value. An absolute value of a difference value between the first difference value and the first preset value is less than an absolute value of a difference value between the second difference value and the second preset value.

Abstract: The present application relates to an atomization device (100) and a counter module (114) for the same. The atomization device (100) comprises: an upper housing (102) and a lower housing (104), wherein the upper housing (102) is configured to mount a container (106) for containing liquid, and the container (106) is movable in an axial direction of the atomization device (100) between a release position and a stretch position, and wherein the stretch position is closer to a proximal end of the atomization device (100) than the release position; and the counter module (114) mounted inside the lower housing (104) and for counting the movement of the container (106) between the release position and the stretch position.

Abstract: A method, system, and computer program product for managing model updates at multiple data centers hosting a same machine learning model obtain a plurality of first feature profiles input to a first implementation of a first machine learning model and a plurality of first model states determined from processing a model input with the first implementation; determine that a first model policy associated with the first machine learning model is satisfied, based on the plurality of first model states including a set of model states defined by the first model policy and the plurality of first feature profiles including a set of feature profiles defined by the first model policy; and in response to determining that the first model policy is satisfied, provide the plurality of first model states and the plurality of first feature profiles for updating at least one second implementation of the first machine learning model.

Abstract: Embodiments of the invention are directed to systems and methods for utilizing a multi-tiered caching architecture in a multi-tenant caching system. A portion of the in-memory cache may be allocated as dedicated shares (e.g., dedicated allocations) that are each dedicated to a particular tenant, while another portion of the in-memory cache (e.g., a shared allocation) can be shared by all tenants in the system. When a threshold period of time has elapsed since data stored in a dedicated allocation has last been accessed, the data may be migrated to the shared allocation. If data is accessed from the shared allocation, it may be migrated back to the dedicated allocation. Utilizing the techniques for providing a multi-tiered approach to a multi-tenant caching system can increase performance and decrease latency with respect to conventional caching systems.