Abstract: A method of categorizing text entries on a document can include determining, for each of a plurality of text bounding boxes in the document, respective text, respective coordinates, and respective input embeddings. The method may further include defining a graph of the plurality of bounding boxes, the graph comprising a plurality of connections among the plurality of bounding boxes, each connection comprising a first and second bounding box and zero or more respective intermediate bounding boxes. The method may further include determining a respective attention value for each connection according to a quantity of intermediate bounding boxes in the connection and, based on a the respective attention values and a transformer-based machine learning model applied to the respective input embeddings and respective coordinates, determining output embeddings for each bounding box and, based on the respective output embeddings, generating a bounding box label for each bounding box.

Abstract: The present disclosure relates to priority flow control (PFC) storm detection and processing methods. In one example method, a first network node performs PFC detection on a first port queue of a first port, and determines that a first preset condition is met. The first preset condition includes: detection is performed in N consecutive first time segments, when the detection is performed in each first time segment, a quantity of first PFC frames sent by the first port queue to a second network node is greater than a first threshold, and a quantity of one or more data packets received by the first port queue from the second network node is less than a second threshold. The first PFC frame is used to indicate the second network node to suspend sending all data flows in the first port queue, and N is a positive integer.

Abstract: Systems and methods for user-agent anomaly detection are disclosed. In one embodiment, a user-agent string may be embedded into a numerical data vector representation using a sentence embedding algorithm (e.g., FastText). A predictive score may be calculated based on the numerical data vector representation and using a probability distribution function model that models a likelihood of occurrence of the observed user-agent based on patterns learned from historic payload data (e.g., a Gaussian Mixture Model). The predictive score may be compared to a threshold and, based on the comparison, it may be determined whether the user-agent is fraudulent.

Abstract: Techniques are disclosed in which a computer system generates a transaction network graph from an initial set of transactions including known labels and attributes. The computer system may generate first and second matrices using first and second graph embedding routines from a training set of transactions that includes a first subset of transactions in the network graph. The first routine is based on anomalies in related transactions occurring at nodes in the transaction network graph that are multiple hops away while the second routine is based on anomalies in neighborhoods of similar transactions. In some embodiments, the computer system generates a final embedded matrix from the first and second matrices and uses the final matrix and a testing set of transactions that includes a second subset of transactions in the graph to train a machine learning model, where the trained model usable to determine whether unlabeled transactions are anomalous.

Abstract: Systems and methods for emulator detection are disclosed. In one embodiment, a user agent string may be embedded into a first numerical data vector representation. Hardware characteristics of a client device corresponding to the user agent may be embedded into a second numerical data vector representation. Based on the first numerical data vector representation of the user agent and the second numerical data vector representation of the hardware characteristics, and their consistency, the client device may be determined to be an emulator or a non-emulator device.

Abstract: Systems and methods are presented for providing a response to a user query. Reception of a user query is detected. An augmentation machine learning model is utilized to determine one or more variations of the user query that correspond to a semantic meaning of the user query. A plurality of response candidates is determined that correspond to the user query by comparing the user query and the one or more variations of the user query to a plurality of documents. A final response candidate is determined from the plurality of response candidates based on utilizing a semantic machine learning model to perform a semantic comparison between the plurality of response candidates and at least the user query.

Abstract: Techniques are disclosed relating to determining whether to authorize a requested action based on whether an entity is an automated computer. In some embodiments, a computer system tracks, at a user interface of a computing device, a sequence of pointer movements. The computer system maps, based on coordinate locations of pointer movements in the sequence, respective movements in the sequence to a plurality of functional areas. Based on the mapping, the computer system generates a movement graph and determines, based on the movement graph, whether an entity associated with the sequence of pointer movements is an automated computer. In response to receiving a request to authorize an action at the computing device, the computer system generates, based on the determining, an authorization decision for the action and transmits the authorization decision to the computing device. Determining whether the entity is an automated computer may advantageously prevent fraudulent activity.

Abstract: Methods and systems are presented for extracting categorizable information from an image using a graph that models data within the image. Upon receiving an image, a data extraction system identifies characters in the image. The data extraction system then generates bounding boxes that enclose adjacent characters that are related to each other in the image. The data extraction system also creates connections between the bounding boxes based on locations of the bounding boxes. A graph is generated based on the bounding boxes and the connections such that the graph can accurately represent the data in the image. The graph is provided to a graph neural network that is configured to analyze the graph and produce an output. The data extraction system may categorize the data in the image based on the output.

Abstract: Methods and systems are presented for generating a device fingerprint based on data obtained from one or more sensors on a device. A plurality of data points corresponding to sensor readings are obtained from the one or more sensors on the device. A set of time-domain features and a set of frequency-domain features are extracted from the plurality of data points and inputted to a neural network trained using a triplet network. A device fingerprint that may be used to identify the device is obtained from the neural network.

Abstract: Systems/techniques for facilitating context-enhanced category classification are provided. In various embodiments, a system can access a first textual description of a product or service. In various aspects, the system can identify, via execution of named entity recognition, one or more keywords in the first textual description. In various instances, the system can access, from a set of queryable databases, one or more second textual descriptions that respectively correspond to the one or more keywords. In various cases, the system can generate, via execution of word embedding, a first numerical representation of the first textual description and one or more second numerical representations of the one or more second textual descriptions. In various aspects, the system can identify, via execution of a machine learning classifier, a category label for the product or service, based on the first numerical representation and the one or more second numerical representations.

Abstract: A system can comprise a processor that can facilitate performance of operations, comprising accessing a document comprising a plurality of text bounding boxes, wherein each respective text bounding box of the plurality of text bounding boxes comprises respective text, for each respective text bounding box, determining respective text bounding box coordinates and respective text bounding box input embeddings, based on the respective text bounding box coordinates, determining respective text bounding box positional encodings for each respective text bounding box, based on a transformer-based deep learning model applied to the respective text bounding box input embeddings, respective text bounding box coordinates, respective text bounding box positional encodings, and bias information representative of a modification to an attention weight of the transformer-based deep learning model, determining respective output embeddings for each respective text bounding box, and based on the respective output embeddings, ge

Abstract: Methods and systems are presented for extracting categorizable information from an image using a graph that models data within the image. Upon receiving an image, a data extraction system identifies characters in the image. The data extraction system then generates bounding boxes that enclose adjacent characters that are related to each other in the image. The data extraction system also creates connections between the bounding boxes based on locations of the bounding boxes. A graph is generated based on the bounding boxes and the connections such that the graph can accurately represent the data in the image. The graph is provided to a graph neural network that is configured to analyze the graph and produce an output. The data extraction system may categorize the data in the image based on the output.

Abstract: Systems and methods for emulator detection are disclosed. In one embodiment, a user agent string may be embedded into a first numerical data vector representation. Hardware characteristics of a client device corresponding to the user agent may be embedded into a second numerical data vector representation. Based on the first numerical data vector representation of the user agent and the second numerical data vector representation of the hardware characteristics, and their consistency, the client device may be determined to be an emulator or a non-emulator device.

Abstract: Systems and methods for user-agent anomaly detection are disclosed. In one embodiment, a user-agent string may be embedded into a numerical data vector representation using a sentence embedding algorithm (e.g., FastText). A predictive score may be calculated based on the numerical data vector representation and using a probability distribution function model that models a likelihood of occurrence of the observed user-agent based on patterns learned from historic payload data (e.g., a Gaussian Mixture Model). The predictive score may be compared to a threshold and, based on the comparison, it may be determined whether the user-agent is fraudulent.

Abstract: Techniques are disclosed in which a computer system generates a transaction network graph from an initial set of transactions including known labels and attributes. The computer system may generate first and second matrices using first and second graph embedding routines from a training set of transactions that includes a first subset of transactions in the network graph. The first routine is based on anomalies in related transactions occurring at nodes in the transaction network graph that are multiple hops away while the second routine is based on anomalies in neighborhoods of similar transactions. In some embodiments, the computer system generates a final embedded matrix from the first and second matrices and uses the final matrix and a testing set of transactions that includes a second subset of transactions in the graph to train a machine learning model, where the trained model usable to determine whether unlabeled transactions are anomalous.

Abstract: Disclosed is an alternating electric field based plane 2D time-grating displacement sensor which includes two parts: the base of fixed ruler and the base of moving ruler, and the two parts are installed in parallel. The base of fixed ruler is arranged with square excitation electrodes that are respectively staggered coded along axis X and Y, and the base of moving ruler is arranged with induction electrodes arranged adjacently along X-axis and Y-axis, and the four induction groups respectively output four traveling wave signals through electric field coupling, and summation of adjacent output signals with an adder can simultaneously decouple two traveling wave signals with opposite phases and only containing X-axis displacement and two traveling wave signals with opposite phases and only containing Y-axis displacement; two traveling wave signals in either direction are differenced by a subtractor to eliminate common mode interference.

Abstract: Methods and systems are presented for extracting categorizable information from an image using a graph that models data within the image. Upon receiving an image, a data extraction system identifies characters in the image. The data extraction system then generates bounding boxes that enclose adjacent characters that are related to each other in the image. The data extraction system also creates connections between the bounding boxes based on locations of the bounding boxes. A graph is generated based on the bounding boxes and the connections such that the graph can accurately represent the data in the image. The graph is provided to a graph neural network that is configured to analyze the graph and produce an output. The data extraction system may categorize the data in the image based on the output.

Abstract: Methods and systems are presented for generating a device fingerprint based on data obtained from one or more sensors on a device. A plurality of data points corresponding to sensor readings are obtained from the one or more sensors on the device. A set of time-domain features and a set of frequency-domain features are extracted from the plurality of data points and inputted to a neural network trained using a triplet network. A device fingerprint that may be used to identify the device is obtained from the neural network.

Abstract: Disclosed is an alternating electric field based plane 2D time-grating displacement sensor which includes two parts: the base of fixed ruler and the base of moving ruler, and the two parts are installed in parallel. The base of fixed ruler is arranged with square excitation electrodes that are respectively staggered coded along axis X and Y, and the base of moving ruler is arranged with induction electrodes arranged adjacently along X-axis and Y-axis, and the four induction groups respectively output four traveling wave signals through electric field coupling, and summation of adjacent output signals with an adder can simultaneously decouple two traveling wave signals with opposite phases and only containing X-axis displacement and two traveling wave signals with opposite phases and only containing Y-axis displacement; two traveling wave signals in either direction are differenced by a subtractor to eliminate common mode interference.

Abstract: Systems and methods are presented for providing a response to a user query. Reception of a user query is detected. An augmentation machine learning model is utilized to determine one or more variations of the user query that correspond to a semantic meaning of the user query. A plurality of response candidates is determined that correspond to the user query by comparing the user query and the one or more variations of the user query to a plurality of documents. A final response candidate is determined from the plurality of response candidates based on utilizing a semantic machine learning model to perform a semantic comparison between the plurality of response candidates and at least the user query.