Abstract: There are provided systems and methods for a quantitative analysis tool for inference speed and performance of AI models. An online transaction processor or other service provider may provide computing services and platforms to entities, which may include services that utilize AI models including LLMs. To provide more efficient training and testing of AI models, a service provider may utilize a quantitative analysis tool that executes operations for algorithmic techniques utilized to calculate model throughput parameters include prefill latency, decoding latency, and other performance metrics. For LLMs, these metrics may be used to compute a time-to-first-token, which may be used to assess an inferencing speed. The tool may analyze the performance metrics and model throughput parameters using model configurations and hardware specifications and may do so without real testing so that system resource usage may be reduced, and performance may be more quickly assessed.

Abstract: Computer security improvements relating to fraud detection and data correlations through large-scale graph clustering of graph transformations and embeddings are disclosed. A service provider may utilize a framework having computing operations for detecting fraud and other malicious or suspicious activities by groups of accounts and fraudsters. In this regard, the service provider may transform relationship graphs of account networks and relationships between accounts and account data captured in the nodes and edges of such graphs. The service provider may merge nodes that edges connecting to other nodes of a certain type of account data, while other types of account data and nodes may not be merged. Edges may also be merged and weighted, and the resulting transformed graph may undergo graph embedding to generate vectors that may be clustered using an AI clustering algorithm. The clusters may then be used for AI model training and inferencing.

Abstract: A system based upon artificial neural networks generates attenuation-corrected SPECT from non-attenuation-corrected SPECT (single photon emission computed tomography) without or with an intermediate step of attenuation map estimation. The system includes a SPECT scanner with CZT cameras for dynamic SPECT imaging. The system also includes a machine learning system including a 3D Dual Squeeze-and-Excitation Residual Dense Network for generating attenuation-corrected SPECT or attenuation maps from non-attenuation-corrected SPECT. The machine learning system reconstructs images from photopeak window and one or more scatter windows of the SPECT scanner are fed to the 3D Dual Squeeze-and-Excitation Residual Dense Network to generate attenuation-corrected SPECT or attenuation maps.

Abstract: A system based upon artificial neural networks generates attenuation-corrected SPECT from non-attenuation-corrected SPECT (single photon emission computed tomography) without or with an intermediate step of attenuation map estimation. The system includes a SPECT scanner with CZT cameras for dynamic SPECT imaging. The system also includes a machine learning system including a 3D Dual Squeeze-and-Excitation Residual Dense Network for generating attenuation-corrected SPECT or attenuation maps from non-attenuation-corrected SPECT. The machine learning system reconstructs images from photopeak window and one or more scatter windows of the SPECT scanner are fed to the 3D Dual Squeeze-and-Excitation Residual Dense Network to generate attenuation-corrected SPECT or attenuation maps.

Abstract: The disclosure provides an electronic device including a housing, a light-emitting element and a light diffusion structure. The housing includes a first light-transmitting hole and a second light-transmitting hole. The light-emitting element is disposed in the housing, and is aligned with the first light-transmitting hole. The light diffusion structure includes a first light diffusion layer and a second light diffusion layer. The first light diffusion layer is disposed between the housing and the light-emitting element, and comprises a first semi-transparent region and a first light-shielding region. The first semi-transparent region corresponds to the first light-transmitting hole, and the first light-shielding region surrounds the first semi-transparent region. The second light diffusion layer is disposed between the housing and the light-emitting element, and comprises a second semi-transparent region and a second light-shielding region.

Abstract: Provided herein are methods of generating positron emission tomography (PET) images from magnetic resonance imaging (MRI) data. In some aspects, the methods can be used for subjects having a condition corresponding to a biomarker of interest, such as a biomarker associated with a brain disorder.

Abstract: A method for detecting a structural abnormal area of a coal and rock mass is provided. The method includes acquiring a target velocity inversion map of the coal and rock mass in a target detection cycle, in which the target velocity inversion map is configured to indicate a target velocity of seismic wave propagation in each of grid cells in a target detection area, acquiring a first velocity of seismic wave propagation in any one of the grid cells in at least one detection cycle preceding the target detection cycle, determining, according to each of the first velocities and the target velocities, structural abnormal cells from each of the grid cells, and determining, according to the structural abnormal cells, a range of the structural abnormal area of the coal and rock mass in the target detection cycle in the target velocity inversion map.

Abstract: A system for estimating attenuation coefficients from only single photon emission computed tomography (SPECT) emission data using deep neural networks includes an artificial neural network based upon machine learning system estimating attenuation maps for SPECT emission data, and associated attenuation correction method.

Abstract: An antenna system includes an antenna body having one or more connection points between the antenna body and a ground; a tunable component having a changeable operating parameter that is a capacitance or a inductance, wherein the antenna body is connected to the tunable component through a first connection point; and a first filter, wherein the first filter are connected to the antenna body; wherein the first filter is connected in parallel with the tunable component, wherein a first end of the first filter is connected to the antenna body through the first connection point, and a second end of the first filter is connected to the ground, and wherein the first filter presents a high impedance characteristic in a low frequency band, and presents a low impedance characteristic in a high frequency band.

Abstract: Embodiments of the present disclosure provide a container image processing method and apparatus, and a storage medium. The method includes: obtaining required file information for generating a container image, and constructing a complete image according to the required file information; determining a specified directory to be deleted in the complete image, and compressing the complete image based on the specified directory to be deleted to obtain a compressed image; obtaining directory content of the specified directory to be deleted, and generating content to be mounted based on the directory content of the specified directory to be deleted; and storing the content to be mounted to a block device file system, and mounting the compressed image to the block device file system.

Abstract: An autonomous mobile device (AMD) follows a user who may move among other people, change orientation, or move in and out view of a camera on the AMD. A predicted location of the user may be determined based on a previous location and previous movement. A proximity value is determined by the distance between a location in physical space of a user and the predicted location. An image from a camera is processed to determine a user depicted in the image and generate a feature vector (FV) of that depiction. A gallery of FVs of the user as viewed from different angles, poses, and so forth is stored. A similarity value is determined between a FV of an unidentified user in an image and the FVs in the gallery. The user may be identified using the proximity value and the similarity value.

Abstract: A system based upon artificial neural networks generates attenuation-corrected SPECT from non-attenuation-corrected SPECT (single photon emission computed tomography) without or with an intermediate step of attenuation map estimation. The system includes a SPECT scanner with CZT cameras for dynamic SPECT imaging. The system also includes a machine learning system including a 3D Dual Squeeze-and-Excitation Residual Dense Network for generating attenuation-corrected SPECT or attenuation maps from non-attenuation-corrected SPECT. The machine learning system reconstructs images from photopeak window and one or more scatter windows of the SPECT scanner are fed to the 3D Dual Squeeze-and-Excitation Residual Dense Network to generate attenuation-corrected SPECT or attenuation maps.

Abstract: The disclosure provides an electronic device including a housing, a light-emitting element and a light diffusion structure. The housing includes a first light-transmitting hole and a second light-transmitting hole. The light-emitting element is disposed in the housing, and is aligned with the first light-transmitting hole. The light diffusion structure includes a first light diffusion layer and a second light diffusion layer. The first light diffusion layer is disposed between the housing and the light-emitting element, and comprises a first semi-transparent region and a first light-shielding region. The first semi-transparent region corresponds to the first light-transmitting hole, and the first light-shielding region surrounds the first semi-transparent region. The second light diffusion layer is disposed between the housing and the light-emitting element, and comprises a second semi-transparent region and a second light-shielding region.

Abstract: Aspects of the invention include a computer-implemented method that receives, by a processor, an ensemble decision tree and generates, by the processor, native code from the ensemble decision tree. The method compiles, by the processor, the native code into machine language and scores, by the processor, the execution time of the native code. The method dynamically reoptimizes, by the processor, portions of the native code corresponding to the most traversed portion of the ensemble decision tree.

Abstract: A slot antenna comprises a printed circuit board, wherein the printed circuit board has a slot, a first capacitor, a radio frequency signal source, a transmission line, and a ground cable, wherein one end of the slot is open, and the other end is closed; the first capacitor and the ground cable are disposed on the printed circuit board, wherein the first capacitor is located on the open end of the slot; the transmission line connects the first capacitor to the radio frequency signal source; and the radio frequency signal source is configured to stimulate a feeding signal, and feed the feeding signal to the open end of the slot, wherein the radio frequency signal source connects the transmission line to the ground cable, and wherein a feed point from the radio frequency signal source to the transmission line is located outside the slot.

Abstract: A system for estimating attenuation coefficients from only single photon emission computed tomography (SPECT) emission data using deep neural networks includes an artificial neural network based upon machine learning system estimating attenuation maps for SPECT emission data, and associated attenuation correction method.

Abstract: Aspects of the invention include a computer-implemented method that receives, by a processor, an ensemble decision tree and generates, by the processor, native code from the ensemble decision tree. The method compiles, by the processor, the native code into machine language and scores, by the processor, the execution time of the native code. The method dynamically reoptimizes, by the processor, portions of the native code corresponding to the most traversed portion of the ensemble decision tree.

Abstract: Systems, methods, and computer-readable media for generating group-based sequential point-of-interest (POI) recommendations are disclosed. A primitive graph or sparse transition matrix is constructed that represents relationships between users, between POIs, and between users and POIs. A loss data prediction algorithm is executed on the primitive graph to obtain a filled graph, which may be a filled user-POI bipartite graph. A group sequential recommendation engine then performs an iterative sequential recommendation process that involves performing, at each iteration, various manipulations on the filled graph to obtain first a homogenous node type graph and then a minimum spanning tree. A next POI to recommend in a sequence of POIs is then determined from the minimum spanning tree. The iterative process continues until stopping criteria is met.

Abstract: The electronic device provided includes a housing with two first through holes and two second through holes, a light guiding structure, and a light source. A first section is provided between the two first through holes, and a second section is provided between the two second through holes. The length of the first section is less than the length of the second section. The light guiding structure includes a light guiding layer, a diffusion layer, and a light adjusting layer. The light adjusting layer is disposed at the position of the first section. The light source is disposed at a position that corresponds to an intersection of the first section and the second section. Light emitted by the light source passes through the light transmitting portion and the intersection, and is separately emitted through the first through holes and the second through holes along the light transmitting portion.

Abstract: The electronic device provided includes a housing with two first through holes and two second through holes, a light guiding structure, and a light source. A first section is provided between the two first through holes, and a second section is provided between the two second through holes. The length of the first section is less than the length of the second section. The light guiding structure includes a light guiding layer, a diffusion layer, and a light adjusting layer. . The light adjusting layer is disposed at the position of the first section. The light source is disposed at a position that corresponds to an intersection of the first section and the second section. Light emitted by the light source passes through the light transmitting portion and the intersection, and is separately emitted through the first through holes and the second through holes along the light transmitting portion.