Abstract: Methods and computing apparatus are provided for training AI/ML systems and use of such systems for performing image analysis so that the damaged parts of a physical structure can be identified accurately and efficiently. According to one embodiment, a method includes selecting an AI/ML system of a particular type; and training the AI/ML system using a dataset comprising one or more auto-labeled images. The auto-labeling was performed using the selected AI/ML system configured using a parts-identification model. The configuration of the trained AI/ML system is output as an improved parts-identification model.

Abstract: The present disclosure relates to systems and methods for using an artificial intelligence technique for determining a source score based on stretched normalization. A natural language query can be received and mapped. Sources can be identified, and actions can be taken with respect to each source. The actions can include determining an item-source metric, transforming the item-source metric using a stretched-normalization factor, and generating a source score based on the transformed item-source metric. A response to the natural language query can be generated based on the source score, and the response can be output.

Abstract: A processor may receive a request for a query item may include a plurality of identifying markers, relating to data associated with the query item. A machine learning model, trained to identify similar items according to the plurality of identifying markers, may then process the plurality of identifying markers and provide a list of one or more similar items and respective similarity distances. The processor may access a respective entity profile including one or more scenario scores for each of the similar items. The processor may then calculate an entity score for each respective entity profile using the respective similarity distances and the scenario scores. The processor may then generate an entity list by ranking the respective entities associated with each respective entity profile using the entity score. The processor may then output the entity list to the client device.

Abstract: A network device may establish, via a routing protocol daemon (RPD) of the network device, border gateway protocol (BGP) sockets with peer network devices and may establish a socket between the RPD and a periodic packet management daemon (PPMD) of the network device. The network device may provide file descriptors of the BGP sockets from the RPD to the PPMD, via the socket, and may provide, from the RPD and via the BGP sockets, non-keep alive protocol data units (PDUs) to the peer network devices. The network device may provide, from the PPMD and via the BGP sockets, keep alive PDUs to the peer network devices.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for using machine learning to detect and correct satellite terminal performance limitations. In some implementations, a system retrieves data indicating labels for clusters of network performance anomalies. The system generates a set of training data to train a machine learning model, the set of training data being generated by assigning the labels for the clusters to sets of performance indicators used to generate the clusters. The system trains a machine learning model to predict classifications for communication devices based on input of performance indicators for the communication devices. The system determines a classification for the communication device based on output that the trained machine learning model generates.

Abstract: A packaged deodorant stick product comprising a deodorant stick composition mounted on an axially immobile retaining member, the retaining member being reversibly connected to a holder enabling the deodorant stick product to be held in the human hand, wherein the retaining member comprises arcuate bridge structures separated by holes, the bridge structures arcing into the deodorant stick composition and being embedded therein.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for using machine learning to detect and correct satellite terminal performance limitations. In some implementations, a system retrieves data indicating labels for clusters of network performance anomalies. The system generates a set of training data to train a machine learning model, the set of training data being generated by assigning the labels for the clusters to sets of performance indicators used to generate the clusters. The system trains a machine learning model to predict classifications for communication devices based on input of performance indicators for the communication devices. The system determines a classification for the communication device based on output that the trained machine learning model generates.

Abstract: A network device may execute a master application shared with another network device via a session, and may receive, by a backup application replication kernel, a replicated data object. The backup application replication kernel may provide the replicated data object to a backup application, and may calculate a time delta between when the replicated data object is received and when the replicated data object is consumed by the backup application. The backup application replication kernel may determine whether the time delta exceeds a first threshold or a second threshold, and may generate a session flag based on the time delta exceeding the first threshold or the second threshold. The backup application replication kernel may provide the session flag to a master application replication kernel and to the backup application, and the master application replication kernel may provide details of the session to the master application and the backup application.

Abstract: The disclosure describes techniques that enable detection of memory leaks of software executing on devices within a computer network. An example network device includes memory and processing circuitry. The processing circuitry monitors a usage of the memory by a software component operating within the network device. The processing circuitry periodically determines a memory growth pattern score for the software component based on the usage of the memory. The processing circuitry also predicts whether the user-level process is experiencing a memory leak based on the memory growth pattern score. The processing circuitry applies confirmation criteria to current memory usage of the software component to confirm that the software component is experiencing the memory leak. When the software component is experiencing the memory leak, the processing circuitry generates an alert.

Abstract: A system and method for collecting, at a terminal, statistical data regarding performance of a transport between the terminal and a satellite; analyzing, at the terminal with an ML model, the statistical data to determine a failure state of the terminal and notifying a user of the failure state, wherein the ML model is trained on a training dataset collected from a terminal population, and the statistical data comprises a terminal state and transport metrics for the terminal.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for using machine learning to detect and correct satellite terminal performance limitations. In some implementations, a system retrieves data indicating labels for clusters of network performance anomalies. The system generates a set of training data to train a machine learning model, the set of training data being generated by assigning the labels for the clusters to sets of performance indicators used to generate the clusters. The system trains a machine learning model to predict classifications for communication devices based on input of performance indicators for the communication devices. The system determines a classification for the communication device based on output that the trained machine learning model generates.

Abstract: A network device may monitor a TCP session with another network device, and may identify ingress and/or egress packets, a TCP header, and a socket of the TCP session. The network device may inspect the ingress and/or egress packets, the TCP header, and the socket to identify a zero window advertisement, details of a last quantity of packets sent or received, synchronize, finish, or reset packets sent or received, negotiated TCP options, or buffer space utilization, and may temporarily record identified data based on the inspection. The network device may detect a TCP session flap when a finish packet or a reset packet is identified and recorded, and may store, in a dead TCP session list, the identified data based on the TCP session flap being detected.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for managing display devices using machine learning. In some implementations, a system receives image data representing an image provided for presentation by a display device. The system processes the image data using a machine learning model that has been trained to evaluate status of display devices based on input of image data corresponding to the display devices. The system selects a classification for a status of the display device based on the output that the machine learning model generated based on the image data. The system provides an output indicating the selected classification over the communication network in response to receiving the image data.

Abstract: A network device may execute a master application communicating with another network device via a session, and may receive, by a backup application replication layer, a replicated data object. The backup application replication layer may provide the replicated data object to a backup application, and may calculate a time delta between when the replicated data object is received and when the replicated data object arrives at the backup application. The backup application replication layer may determine whether the time delta exceeds a first threshold or a second threshold, and may generate a session flag based on the time delta exceeding the first threshold or the second threshold. The backup application replication layer may provide the session flag to a master application replication layer and to the backup application, and the master application replication layer may provide details of the session to the master application and the backup application.

Abstract: Methods and systems for monitoring a communication network using machine-learning techniques are disclosed. In some implementations, a forecasted amount of traffic for a communication network is determined using one or more network traffic forecasting models being configured to generate the forecasted amount of traffic based on data indicating one or more previous amounts of traffic for the communication network. A measure of network health is generated based on a measured amount of traffic and the forecasted amount of traffic. Data indicating one or more characteristics of the communication network is processed using one or more machine learning models to generate a predicted measure of network health for a future time period. An indication of the predicted measure of network health for the future time period is provided.

Abstract: A network device may execute a master application communicating with another network device via a session, and may receive, by a backup application replication layer, a replicated data object. The backup application replication layer may provide the replicated data object to a backup application, and may calculate a time delta between when the replicated data object is received and when the replicated data object arrives at the backup application. The backup application replication layer may determine whether the time delta exceeds a first threshold or a second threshold, and may generate a session flag based on the time delta exceeding the first threshold or the second threshold. The backup application replication layer may provide the session flag to a master application replication layer and to the backup application, and the master application replication layer may provide details of the session to the master application and the backup application.

Abstract: A packaged deodorant stick product comprising a deodorant stick composition mounted on an axially immobile retaining member, the retaining member being reversibly connected to a holder enabling the deodorant stick product to be held in the human hand, wherein the retaining member comprises arcuate bridge structures separated by holes, the bridge structures arcing into the deodorant stick composition and being embedded therein.

Abstract: A processor may receive a request for a query item may include a plurality of identifying markers, relating to data associated with the query item. A machine learning model, trained to identify similar items according to the plurality of identifying markers, may then process the plurality of identifying markers and provide a list of one or more similar items and respective similarity distances. The processor may access a respective entity profile including one or more scenario scores for each of the similar items. The processor may then calculate an entity score for each respective entity profile using the respective similarity distances and the scenario scores. The processor may then generate an entity list by ranking the respective entities associated with each respective entity profile using the entity score. The processor may then output the entity list to the client device.

Abstract: An automated automobile claims fraud detector and method for automatically evaluating validity and extent of at least one damaged object from image data and detect possible fraud, the automated automobile claims fraud detector comprising the processing steps of: (a) receiving image data comprising one or more images of at least one damaged object; (b) processing said one or more images for existing image alteration using unusual pattern identification and providing a first fraud detection; (c) processing said one or more images for fraud detection using RGB image input, wherein the RGB values are used for (i) CNN-based pre-existing damage detection, (ii) parallel CNN-based color matching, and (iii) double JPEG compression (DJCD) detection using custom CNN; (d) processing output of (i) CNN-based pre-existing damage detection, (ii) parallel CNN-based color matching, and (iii) double JPEG compression detection using custom CNN as input for ML-based fraud identifier providing a second fraud detection; and (e) gen

Abstract: A method and network device for detecting a TCP socket failure is described. A network device may be configured to detect a failure of a link between a first node and a second node, determine one or more transmission control protocol (TCP) sockets of a plurality of TCP sockets on the first node that are communicating over the link between the first node and the second node, write information to a TCP stack for the determined one or more TCP sockets, the information indicating that the determined one or more TCP sockets have an error, and remediate the determined one or more TCP sockets in response to the information.