Abstract: A multi-instance learning and weakly supervised BGP anomaly detection framework is provided, that detects and analyzes significant statistical correlations across multiple data sources such as model driven telemetry (MDT), network messages, event data logs, and/or device configuration data for network topology. Specifically, methods are provided that involve obtaining, from a plurality of data sources, data related to operation or configuration of Border Gateway Protocol (BGP) in an enterprise network and extracting one or more BGP features based on at least one correlation among the data from the plurality of data sources. The methods further involve detecting one or more network anomalies by performing a weakly supervised machine learning of the one or more BGP features and providing information about the one or more network anomalies for performing one or more actions associated with the enterprise network.

Abstract: Large Language Model (LLM) driven proactive scheduling may be provided. First, a proactive feedback module may be used that gathers user requests and device feedback. Next, an instructive interpreter module may be used that receives the user requests and the device feedback and produces instructive prompts based on the user requests and the device feedback. Then a user-reinforced scheduling optimization module may be used that receives responses to the instructive prompts and continuously enhances bandwidth scheduling based on the receives responses.

Abstract: A method, computer system, and computer program product are provided for generating personalized knowledge session content using a generative machine learning model. User profile data of a plurality of users and service description data that is descriptive of a service are processed to assign the plurality of users to a plurality of user groups. A personalized message is generated for each user group based on the user profile data and the service description data, wherein the personalized message is generated using one or more machine learning models. Each personalized message is provided to each user group and feedback data is collected based on the providing of each personalized message. Based on analyzing of the feedback data, one or more of: the user profile data, the one or more machine learning models, and the service description data are updated.

Abstract: Methods are provided for generating digests of network-related notifications specifically tailored to user's personas and adaptable across multiple timescale frequencies. Specifically, the methods involve obtaining user data of a user associated with an enterprise network and a plurality of network-related notifications. Each of the plurality of network-related notifications relates to network operations or network configurations. The methods further involve determining a network persona of the user in a context of the enterprise network based on the user data and generating a digest of the plurality of network-related notifications based on the network persona. The digest includes a semantic summary for each of the plurality of network-related notifications that is specific to the network persona. The methods further involve providing the digest for performing one or more actions associated with the enterprise network.

Abstract: The present disclosure describes a network that uses artificial intelligence to set a channel access protocol. A network device includes one or more memories and one or more processors communicatively coupled to the one or more memories. The one or more processors, individually or collectively predict, using a first machine learning model, an idle time for a wireless medium for a future period of time, predict, using a second machine learning model and based on the predicted idle time, a minimum contention window value and a maximum contention window value for the wireless medium, and wirelessly transmit an instruction to use the minimum contention window value and the maximum contention window value when communicating over the wireless medium during the period of time.

Abstract: Methods are provided for generating hierarchical summaries with actionable recommendations having various granularities. Specifically, the methods involve obtaining notifications related to network issues and generating meta-semantic data that includes a summary of each of the notifications. The methods further involve obtaining inventory data of network devices in a plurality of domains of a network. The inventory data includes configuration information of the network devices. The methods further involve generating a multi-level hierarchical summary specific to the network based on the inventory data and the meta-semantic data. The multi-level hierarchical summary includes a first level specific to one or more affected network devices and a second level specific to a group of network devices. The methods further involve providing the multi-level hierarchical summary for performing one or more actions associated with the network.

Abstract: A unified prompt-based network management system that involves an intelligent auto-prompt engine generating contextualized prompts for an artificial intelligence model. The artificial intelligence model generates instructions and/or solutions and adapts to different application scenarios based on an enterprise network knowledge and reverse inference(s). Specifically, methods are provided that involve obtaining input data related to a configuration or an operation of one or more assets in an enterprise network and generating a contextualized prompt based on the input data, network knowledge information of the enterprise network, and at least one reverse inference generated using an artificial intelligence model. The methods further involve providing the contextualized prompt to the artificial intelligence model for generating a tailored response to the input data, wherein the tailored response includes a set of actionable tasks to be performed with respect to the one or more assets of the enterprise network.

Abstract: Methods are provided for generating digests of network-related notifications specifically tailored to user's personas and adaptable across multiple timescale frequencies. Specifically, the methods involve obtaining user data of a user associated with an enterprise network and a plurality of network-related notifications. Each of the plurality of network-related notifications relates to network operations or network configurations. The methods further involve determining a network persona of the user in a context of the enterprise network based on the user data and generating a digest of the plurality of network-related notifications based on the network persona. The digest includes a semantic summary for each of the plurality of network-related notifications that is specific to the network persona. The methods further involve providing the digest for performing one or more actions associated with the enterprise network.

Abstract: Methods are provided for generating distilled multimedia data sets tailored to user's persona and/or task(s) to be performed associated with an enterprise network and enable interactive contextual learning using a multi-modal knowledge graph. Methods involve obtaining multimedia data from one or more data sources related to operation or configuration of an enterprise network and determining context for generating a distilled multimedia data set based on at least one of user input and user persona. The methods further involve generating, based on the context, the distilled multimedia data set that includes a set of multimedia slices generated from the multimedia data using a multi-modal knowledge graph. The multi-modal knowledge graph is generated using a graph neural network and indicates relationships among a plurality of slices of the multimedia data. The methods further involve providing the distilled multimedia data set for performing one or more actions associated with the enterprise network.

Abstract: A heterogeneous graph learning system generates and analyzes network implementations. The heterogeneous graph learning system includes obtaining information describing multiple network implementations including heterogeneous nodes. The heterogeneous graph learning system also includes generating a one-hop graph connecting a particular node of the heterogeneous nodes with a set of related nodes. The one-hop graph connects the particular node with the set of related nodes via corresponding edges. The heterogeneous graph learning system further includes transforming the one-hop graph into a weighted graph based on a Dynamic Meta Path Transformation (DMPT). In the DMPT, each of the corresponding edges connecting the particular node to a corresponding related node among the set of related nodes is associated with a corresponding weight.

Abstract: Methods are provided which involve obtaining enterprise data about a plurality of assets and configuration of an enterprise network, and partner data about one or more network related partner services for the enterprise network. The methods further involve determining one or more hierarchical relationships among the plurality of assets, the enterprise network, and the one or more network related partner services, by performing machine learning on the enterprise data and the partner data. Additionally, the methods involve generating one or more risk values based on the one or more hierarchical relationships and providing the one or more risk values indicative of performance of the one or more network related partner services.

Abstract: A heterogeneous graph learning system generates and analyzes network implementations. The heterogeneous graph learning system includes obtaining information describing multiple network implementations including heterogeneous nodes. The heterogeneous graph learning system also includes generating a one-hop graph connecting a particular node of the heterogeneous nodes with a set of related nodes. The one-hop graph connects the particular node with the set of related nodes via corresponding edges. The heterogeneous graph learning system further includes transforming the one-hop graph into a weighted graph based on a Dynamic Meta Path Transformation (DMPT). In the DMPT, each of the corresponding edges connecting the particular node to a corresponding related node among the set of related nodes is associated with a corresponding weight.

Abstract: Methods are provided for generating end-to-end solutions-based search results for multi-query search inquiry. The search results are generated using graph generative pre-trained transformers and a network knowledge base. A method involves obtaining at least one search query and inventory data that includes information about a plurality of enterprise assets and configuration of an enterprise network. The method further includes generating a contextual schema based on the inventory data. The contextual schema includes a plurality of query sub-graphs indicative of an intention of the at least one search query and generating a solution graph by performing machine learning with respect to the plurality of query sub-graphs and network domain knowledge data. The method further includes providing a response to the at least one search query based on the solution graph. The response is specific to the enterprise network.

Abstract: A platform dynamically detects a user persona and facilitates a user objective in a user session. The platform obtains user activity of users across multiple historical sessions. The platform clusters similar user activity across the historical sessions to determine personas being used in the historical sessions. The platform mines sequential patterns in a set of user activity data associated with one of the personas and determines at least one predictive rule associated with that persona. Each predictive rule includes an initial activity among the set of user activity data and at least one subsequent activity in the set of user activity data.

Abstract: A platform dynamically detects a user persona and facilitates a user objective in a user session. The platform obtains user activity of users across multiple historical sessions. The platform clusters similar user activity across the historical sessions to determine personas being used in the historical sessions. The platform mines sequential patterns in a set of user activity data associated with one of the personas and determines at least one predictive rule associated with that persona. Each predictive rule includes an initial activity among the set of user activity data and at least one subsequent activity in the set of user activity data.

Abstract: Methods are provided for generating hierarchical summaries with actionable recommendations having various granularities. Specifically, the methods involve obtaining notifications related to network issues and generating meta-semantic data that includes a summary of each of the notifications. The methods further involve obtaining inventory data of network devices in a plurality of domains of a network. The inventory data includes configuration information of the network devices. The methods further involve generating a multi-level hierarchical summary specific to the network based on the inventory data and the meta-semantic data. The multi-level hierarchical summary includes a first level specific to one or more affected network devices and a second level specific to a group of network devices. The methods further involve providing the multi-level hierarchical summary for performing one or more actions associated with the network.

Abstract: A method, computer system, and computer program product are provided for mitigating network risk. A plurality of risk reports corresponding to a plurality of network devices in a network are processed to determine a multidimensional risk score for the network. The plurality of risk reports are analyzed using a semantic analysis model to identify one or more factors that contribute to the multidimensional risk score. One or more actions are determined using a trained learning model to mitigate one or more dimensions of the multidimensional risk score. The outcomes of applying the one or more actions are presented to a user to indicate an effect of each of the one or more actions on the multidimensional risk score for the network.

Abstract: AP coordination, and more specifically intelligent AP coordination using a graph network and reinforcement learning may be provided. AP coordination may include translating a physical space into a logical space, wherein the physical space is being evaluated for AP coordination. A machine learning process may predict signal strengths of signals sent by one or more Access Points (APs) and received by one or more Stations (STAs), wherein the machine learning process uses the logical space, and wherein each STA is in a location of the physical space. One or more AP placements may be evaluated based on the signal strengths, and a recommended AP placement may be determined based on the evaluation.

Abstract: A platform dynamically detects a user persona and facilitates a user objective in a user session. The platform obtains user activity of users across multiple historical sessions. The platform clusters similar user activity across the historical sessions to determine personas being used in the historical sessions. The platform mines sequential patterns in a set of user activity data associated with one of the personas and determines at least one predictive rule associated with that persona. Each predictive rule includes an initial activity among the set of user activity data and at least one subsequent activity in the set of user activity data.

Abstract: Methods are provided in which a computing device obtains user data and network data associated with one or more assets used in an enterprise network of a user. The computing device further determines an identity of the user based on the user data and the network data and generates a task recommendation based on the identity of the user. The task recommendation includes one or more tasks having a plurality of operations that are to be performed within a predetermined time interval. The computing device further provides the task recommendation for performing one or more actions associated with configuring the enterprise network.