Abstract: Disclosed are an advanced persistent threat (APT) detection method and system based on a continuous-time dynamic heterogeneous graph network (CDHGN). The method includes: selecting network interaction event data, extracting entities from the network interaction event data as source nodes and target nodes, extracting an interaction event occurring between a source node and a target node as an edge, and determining a type and an attribute of a node, a type and an attribute of the edge, and a moment at which an interaction event occurs, to obtain a continuous-time dynamic heterogeneous graph; converting each type of edge in the continuous-time dynamic heterogeneous graph into a vector by a CDHGN encoder, to obtain an embedding representation by a CDHGN decoder to obtain a detection result of whether each type of edge is an abnormal edge, to intercept an APT attack.

Abstract: A method of personalized image retrieval includes obtaining a natural language query including a name; replacing the name in the natural language query with a generic term to provide an anonymized query and named entity information; obtaining a plurality of initial ranking scores and a plurality of attention weights corresponding to a plurality of images using a trained scoring model that inputs the anonymized query and the plurality of images; obtaining a plurality of delta scores corresponding to the plurality of images using a re-scoring model that inputs the plurality of attention weights and the named entity information; and obtaining a plurality of final ranking scores by modifying the plurality of initial ranking scores based on the plurality of delta scores. The trained scoring model performs semantic based searching and the re-scoring model determines a probability that faces detected in the plurality of images correspond to the name.

Abstract: Disclosed is a lightweight two-crossbeam high-speed free forging hydraulic press. The hydraulic press includes a fixed crossbeam, a movable crossbeam, a column, and a return oil cylinder. By integrating a hydraulic cylinder with a through hole for the column to slide, an annular hydraulic cylinder is formed between the movable crossbeam and the column, and no hole is required to be specially provided on the movable crossbeam for mounting the hydraulic cylinder. Therefore, the rigidity, strength and bending resistance of the movable crossbeam serving as an upper crossbeam or a lower crossbeam can be effectively guaranteed, the crossbeam structure is simplified, and the manufacturing difficulty is reduced. Meanwhile, compared with the existing hydraulic press of a three-crossbeam structure, the two-crossbeam hydraulic press can greatly reduce the overall dimension and mass of the high-speed forging hydraulic press, reduce the overall height of the hydraulic press.

Abstract: The present disclosure relates to a lightweight and small bathymetry LiDAR multi-channel high sampling rate high-precision real-time synchronous acquisition and storage system, which adopts an ADC+FPGA+ZYNQ architecture to implement four-channel high-rate real-time synchronous parallel sampling, has synchronization error less than 300 ps, sampling rate s as high as 2 GSPS, and sampling precision as high as 14 bits, and comprises an FPGA system acquisition carrier board unit for implementing laser radar echo data acquisition and storage, PMT controlling, data maximum value feedback, peripheral interface design and storage control; a storage daughter board unit for implementing storage of echo data and export of 100 Mbps Ethernet, and an upper computer data conversion software for implementing conversion of original echo data files into decimal or hexadecimal csv files.

Abstract: Disclosed are a method and system for detecting a complex multi-step attack in an electric power system. The method includes: collecting interaction behavior data of a network entity; preprocessing the interaction behavior data of the network entity based on a heterogeneous graph to obtain input data; and inputting the input data into a complex multi-step attack detection module to obtain an attack detection result. Information is extracted from the interaction behavior data to construct a node and an edge of the heterogeneous graph. Timestamp information of a destination node and an adjacent source node of the destination node is input into a Time2Vec layer to obtain a first time embedding representation. Data that fuses node feature information and the first time embedding representation is input into a Heteformer layer, and a second node embedding representation is obtained as input data and input into the complex multi-step attack detection module.

Abstract: In various examples, artificial intelligence (AI) agents can be generated to synthesize more natural motion by simulated actors in various visualizations (such as video games or simulations). AI agents may employ one or more machine learning models and techniques, such as reinforcement learning, to enable synthesis of motion with enhanced realism. The AI agent can be trained based on widely-available broadcast video data, without the need for more costly and limited motion capture data. To account for the lower quality of such video data, various techniques can be employed, such as taking into account the motion of joints, and applying physics-based constraints on the actors, resulting in higher quality, more lifelike motion.

Abstract: A method of personalized image retrieval includes obtaining a natural language query including a name; replacing the name in the natural language query with a generic term to provide an anonymized query and named entity information; obtaining a plurality of initial ranking scores and a plurality of attention weights corresponding to a plurality of images using a trained scoring model that inputs the anonymized query and the plurality of images; obtaining a plurality of delta scores corresponding to the plurality of images using a re-scoring model that inputs the plurality of attention weights and the named entity information; and obtaining a plurality of final ranking scores by modifying the plurality of initial ranking scores based on the plurality of delta scores. The trained scoring model performs semantic based searching and the re-scoring model determines a probability that faces detected in the plurality of images correspond to the name.

Abstract: The disclosure relates to a gas grill and control method thereof. The gas grill comprises a housing defining a cooking chamber, a plurality of cooking zones in the cooking chamber, wherein each of the plurality of cooking zones is corresponding to at least one heating module, a gas supply module configured to provide gas to the at least one heating module of each of the plurality of cooking zones, a control module configured to control the gas supply module and the at least one heating module corresponding to each of the plurality of cooking zones based on a control input, and a user interface configured to generate the control input, wherein the control module is further configured to group at least two of the plurality of cooking zones to a cooking group based on the control input and control the heating modules corresponding to the cooking group based on the control input.

Abstract: A method of personalized image retrieval includes obtaining a natural language query including a name; replacing the name in the natural language query with a generic term to provide an anonymized query and named entity information; obtaining a plurality of initial ranking scores and a plurality of attention weights corresponding to a plurality of images using a trained scoring model that inputs the anonymized query and the plurality of images; obtaining a plurality of delta scores corresponding to the plurality of images using a re-scoring model that inputs the plurality of attention weights and the named entity information; and obtaining a plurality of final ranking scores by modifying the plurality of initial ranking scores based on the plurality of delta scores. The trained scoring model performs semantic based searching and the re-scoring model determines a probability that faces detected in the plurality of images correspond to the name.

Abstract: The invention discloses an indoor adaptive ventilation system and method based on multi-vent ventilation modules, comprising at least one multi-vent ventilation module, the multi-vent ventilation modules comprise air reversing valves and ventilation ducts, the multi-vent ventilation ducts are connected with the air reversing valves, the air reversing valve comprises a box body, a box cover, a reversing mechanism, a supply air main duct, a return air main duct, a fixed supply air branch duct and multiple variable air direction branch ducts; a plurality of variable air direction branch ducts are evenly arranged in the circumferential direction of the box body. The invention can divide the indoor space into different sub-spaces, control pollutants in a smaller area, reduce the possibility of large-scale pollutant diffusion, and realize that the supply air outlets and the return air outlets can be switched mutually to adapt to different usage scenarios and needs.

Abstract: Disclosed is an aliphatic polyisocyanate composition, wherein the molar ratio of an iminooxadiazinedione structure to an isocyanurate structure is 0.01-0.8:1, and the molar ratio of an iminooxadiazinedione structure in a polymer larger than three molecules to the iminooxadiazinedione structure in the composition is 0.2-0.8:1. Provided are a method for preparing the aliphatic polyisocyanate composition, and a coating composition. The aliphatic polyisocyanate composition has both a lower viscosity and a better crosslinking property. The composition can significantly reduce the usage amount of a solvent when used in a coating composition, is more environmentally friendly and also has a better paint film property. The preparation method has a simple process and mild conditions, is easy to operate and control, and is suitable for industrial scale production.

Abstract: An apparatus for scrubbing an image may include a memory storing instructions; and a processor configured to execute the instructions to: receive an input image; input a preset public attribute to an encoder neural network; obtain a scrubbed feature from the input image based on the preset public attribute, via the encoder neural network; wherein the encoder neural network is trained based on an amount of information in the scrubbed feature about the input image, and an estimated public attribute estimated from the scrubbed feature.

Abstract: A method of personalized image retrieval includes obtaining a natural language query including a name; replacing the name in the natural language query with a generic term to provide an anonymized query and named entity information; obtaining a plurality of initial ranking scores and a plurality of attention weights corresponding to a plurality of images using a trained scoring model that inputs the anonymized query and the plurality of images; obtaining a plurality of delta scores corresponding to the plurality of images using a re-scoring model that inputs the plurality of attention weights and the named entity information; and obtaining a plurality of final ranking scores by modifying the plurality of initial ranking scores based on the plurality of delta scores. The trained scoring model performs semantic based searching and the re-scoring model determines a probability that faces detected in the plurality of images correspond to the name.

Abstract: The disclosure relates to a gas grill and control method thereof. The gas grill comprises a housing defining a cooking chamber, a plurality of cooking zones in the cooking chamber, wherein each of the plurality of cooking zones is corresponding to at least one heating module, a gas supply module configured to provide gas to the at least one heating module of each of the plurality of cooking zones, a control module configured to control the gas supply module and the at least one heating module corresponding to each of the plurality of cooking zones based on a control input, and a user interface configured to generate the control input, wherein the control module is further configured to group at least two of the plurality of cooking zones to a cooking group based on the control input and control the heating modules corresponding to the cooking group based on the control input.

Abstract: Techniques of inpainting video content include training a neural network to perform an inpainting operation on a video using only content from that video. For example, upon receiving video content including a sequence of initial frames, a computer generates a sequence of inputs corresponding to at least some of the sequence of initial frames and each input including, for example, a uniform noise map. The computer then generates a convolutional neural network (CNN) using the sequence of input as the initial layer. The parameters of the CNN are adjusted according to a cost function, which has components including a flow generation loss component and a consistency loss component. The CNN then outputs, on a final layer, estimated image values in a sequence of final frames.

Abstract: An apparatus for scrubbing an image may include a memory storing instructions; and a processor configured to execute the instructions to: receive an input image; input a preset public attribute to an encoder neural network; obtain a scrubbed feature from the input image based on the preset public attribute, via the encoder neural network; wherein the encoder neural network is trained based on an amount of information in the scrubbed feature about the input image, and an estimated public attribute estimated from the scrubbed feature.

Abstract: Techniques of inpainting video content include training a neural network to perform an inpainting operation on a video using only content from that video. For example, upon receiving video content including a sequence of initial frames, a computer generates a sequence of inputs corresponding to at least some of the sequence of initial frames and each input including, for example, a uniform noise map. The computer then generates a convolutional neural network (CNN) using the sequence of input as the initial layer. The parameters of the CNN are adjusted according to a cost function, which has components including a flow generation loss component and a consistency loss component. The CNN then outputs, on a final layer, estimated image values in a sequence of final frames.

Abstract: Techniques for ranking of selected bots are described. In one embodiment, for example, an apparatus may comprise a client front-end component operative to receive a bot contact display prompt from a client device; and send an ordered bot contact list to the client device; a bot contact list component operative to retrieve a bot contact list from a selection component, the bot contact list comprising a plurality of bot contacts; and a contact ranking component operative to determine a ranking weight for each of the plurality of bot contacts; and generate the ordered bot contact list by ordering the bot contact list based on the ranking weight. Other embodiments are described and claimed.

Abstract: Systems and methods are disclosed relating generally to data processing, and more particularly to applying low pass and rotation filtering in relation to data processing. For example, a system may include a phase modification value determination circuit operable to generate a phase offset value based upon an input data set derived from information sensed from a storage medium. The system may include an integrated low pass and rotation filtering circuit operable to simultaneously apply a low pass filtering function and phase rotation function to a series of digital samples derived from the information sensed from the storage medium to yield a modified output. Application of both the low pass filtering and phase rotation functions is governed at least in part based upon a selected coefficient set corresponding to a combination of the phase off set value and a boost value.

Abstract: Systems and methods are disclosed relating generally to data processing, and more particularly to applying low pass and rotation filtering in relation to data processing. For example, a system may include a phase modification value determination circuit operable to generate a phase offset value based upon an input data set derived from information sensed from a storage medium. The system may include an integrated low pass and rotation filtering circuit operable to simultaneously apply a low pass filtering function and phase rotation function to a series of digital samples derived from the information sensed from the storage medium to yield a modified output. Application of both the low pass filtering and phase rotation functions is governed at least in part based upon a selected coefficient set corresponding to a combination of the phase off set value and a boost value.