Abstract: A path computing method, system, and computer program product, include extracting unpleasant data from a database to create a multivariate spatia-temporal density function, collecting a tolerance level of a user, and computing a path for the user based on the tolerance level and the density function.

Abstract: A first feature set for an accident is identified, based on one or more historical accident reports. The accident is geolocated to obtain road data for a location of the accident. Weather data for the accident is obtained, based on the accident location. Hidden data for the accident is obtained, based on the accident location and the time of the accident. The first feature set is combined with the road data, weather data, and hidden data to produce a complete feature set for the accident. A model feature set is generated, based on a combination of the complete feature set and a plurality of other complete feature sets produced based on historical accident reports for a plurality of other accidents.

Abstract: A method, computer system, and a computer program product for identifying feature importance in deep learning models is provided. Embodiments of the present invention may include building a reconstruction model. Embodiments of the present invention may include intercepting an output of a trained prediction model at a bottleneck layer. Embodiments of the present invention may include processing the output of the trained model using the reconstruction model. Embodiments of the present invention may include identifying a plurality of features based on the reconstruction model.

Abstract: A computer-implemented method includes: receiving, using a processor, multiple data session records (DSRs); storing the multiple DSRs in a memory communicatively coupled to the processor; analyzing, using the processor, the stored multiple DSRs for temporal and spatial data; and determining, using the processor, quality degradation by using the temporal and spatial data for the stored multiple DSRs.

Abstract: Data associated with performances of microservices functioning in a distributed computing environment is clustered by executing an unsupervised machine learning algorithm. A representative data is selected from a cluster, selecting performed for a plurality of the clusters. Based on time series data of the representative data associated with the plurality of the clusters, causal extraction is performed. Based on the causal extraction and the plurality of the clusters, a causal graph is constructed. The causal graph is embedded into vector space. Based on the embedded vector space, an artificial neural network model can be trained for managing the distributed computing environment.

Abstract: Systems and techniques are disclosed for using machine learning to dynamically detect physical impairments in lines of a subscriber network. In some implementations, per-tone data for a line of a subscriber network and data indicating a set of one or more scores is obtained. Each score included in the set of scores indicates a conditional likelihood that the line has a type of impairment with respect to a different feature subset ensemble. The per-tone data and the data indicating the set of one or more scores is provided as input to a model. The model is trained to output, for each of different sets of feature subset ensembles, a confidence score representing an overall likelihood that a particular line has a physical impairment. Data indicating a particular confidence score representing an overall likelihood that the line has the physical impairment is obtained. The particular confidence score is provided for output.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for telecommunications network troubleshooting. In one aspect, a method includes obtaining multiple current observations characterizing current operation of a telecommunications network, the multiple observations including performance monitoring data for the network and quality of service data for the network. One or more prior network states for the network are obtained. A current state of the network is determined, by a machine learning model, based on the current observations and the one or more prior network states. One or more actions are performed based on the current state of the network, including, when the current state of the network is an abnormal state, taking an action that is mapped to the abnormal state.

Abstract: A method, system and computer program product for adaptive system monitoring. In one embodiment, the method comprises generating time-varying correlation graphs indicating correlations between a multitude of parameters of the system, and using the correlation graphs to identify monitoring logic for monitoring the system. In an embodiment, the correlation graphs are used to select a group of the parameters as monitoring parameters, and these monitoring parameters are dynamically changed. In one embodiment, the monitoring parameters form sets of monitoring parameters, and each set of monitoring parameters is used to monitor the system for an associated period of time. The lengths of these monitoring periods are changed based on the rate of change of the correlation graphs. In an embodiment, the rate at which the monitoring parameters are changed is itself changed based on the rate of change of the correlation graphs.

Abstract: A method, system and computer program product for transmitting content across a link in a communications network. In an embodiment, the method comprises examining traffic in the network at a sender end of the link to determine if the traffic is carrying compressed content; when the traffic is carrying compressed content, decompressing at least one portion of the compressed content; and checking a cache to determine if the cache has content matching the at least one decompressed portion of the content. In an embodiment, the method further comprises when the cache has content matching the at least one decompressed portion of the content, determining an identifier associated with the at least one decompressed portion of the content; and using the identifier to reconstruct, at a receiving end of the link, the at least one decompressed portion of the content.

Abstract: A telecommunication system uses a dynamic bandwidth allocation (DBA) algorithm based on current load conditions for controlling transmissions to a plurality of access modules of an access node in order to achieve a fair allocation of network bandwidth at the access node. As an example, access modules at an access node communicate via a control channel with dynamic bandwidth allocation (DBA) logic that receives load information from each of the access modules. Using such load information, the DBA logic dynamically controls the upstream data rates so that a fair allocation of network bandwidth is achieved across all of the access modules. Specifically, the data rates are controlled such that packet flows for services of the same class achieve the same or similar performance (e.g., average data rate) regardless of which access module is receiving each respective packet flow.

Abstract: A method, system and computer program product for adaptive system monitoring. In one embodiment, the method comprises generating time-varying correlation graphs indicating correlations between a multitude of parameters of the system, and using the correlation graphs to identify monitoring logic for monitoring the system. In an embodiment, the correlation graphs are used to select a group of the parameters as monitoring parameters, and these monitoring parameters are dynamically changed. In one embodiment, the monitoring parameters form sets of monitoring parameters, and each set of monitoring parameters is used to monitor the system for an associated period of time. The lengths of these monitoring periods are changed based on the rate of change of the correlation graphs. In an embodiment, the rate at which the monitoring parameters are changed is itself changed based on the rate of change of the correlation graphs.

Abstract: A request may be received to join one or more attributes of at least two independent sets of data into a data structure. The one or more attributes may include a time attribute. The two independent sets of data may be included within a data store. It may be determined that there are one or more null values associated with the join to the data structure. In response to at least the determining that there are one or more null values associated with the join, one or more values may be imputed into one or more fields corresponding to the one or more null values, wherein there are no null values in the one or more fields subsequent to the imputing.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for an artificial intelligence system. In one aspect, a system includes multiple artificial intelligence skill agents that have each been trained to perform different actions in a telecommunications system. The system also includes an orchestrator agent that interacts with each of the artificial intelligence skill agents and coordinates which of the artificial intelligence agents performs actions in response to user inputs. The orchestrator agent receives a user input and determines an intent expressed by the user input. The orchestrator agent transmits an instruction to an artificial intelligence skill agent to perform an action that provides a response to the intent. In response to receiving the instruction from the orchestrator agent, the artificial intelligence skill agent performs the action when the artificial intelligence skill agent is capable of carrying out the action.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for telecommunications network troubleshooting. In one aspect, a method includes obtaining multiple current observations characterizing current operation of a telecommunications network, the multiple observations including performance monitoring data for the network and quality of service data for the network. One or more prior network states for the network are obtained. A current state of the network is determined, by a machine learning model, based on the current observations and the one or more prior network states. One or more actions are performed based on the current state of the network, including, when the current state of the network is an abnormal state, taking an action that is mapped to the abnormal state.

Abstract: A request may be received to join one or more attributes of at least two independent sets of data into a data structure. The one or more attributes may include a time attribute. The two independent sets of data may be included within a data store. It may be determined that there are one or more null values associated with the join to the data structure. In response to at least the determining that there are one or more null values associated with the join, one or more values may be imputed into one or more fields corresponding to the one or more null values, wherein there are no null values in the one or more fields subsequent to the imputing.

Abstract: A method, computer system, and computer program product for a cognitive, real-time feedback speaking coach are provided. The embodiment may include capturing a plurality of text from a prepared document. The embodiment may also include capturing a plurality of user voice data and a plurality of user movement data. The embodiment may further include calculating a speaker rating based on the plurality of received user voice data, the plurality of received user movement data, and the plurality of captured text. The embodiment may also include identifying one or more points of improvement based on the calculated speaker rating. The embodiment may further include alerting a user of the one or more identified points of improvement.

Abstract: A path computing method, system, and computer program product, include extracting unpleasant data from a database to create a multivariate spatia-temporal density function, collecting a tolerance level of a user, and computing a path for the user based on the tolerance level and the density function.

Abstract: Techniques to adaptively launch/replace applications and services on edge devices in a cellular infrastructure and/or adaptively place content and computation at the edge devices based on logic at the network core are provided. In one aspect, a method for dynamic placement of applications in a cellular network mobile cloud is provided which includes the steps of: (a) obtaining: (i) a model of the cellular network, (ii) a model of user mobility patterns in the cellular network, and (iii) a model of a profile of the applications; (b) upon receipt of requests from users for the applications, obtaining runtime states at edge servers in the mobile cloud; and (c) upon obtaining the runtime states at the edge servers, placing the requests among the edge servers and a core server in the mobile cloud based on the models obtained in step (a) and the runtime states obtained in step (b).

Abstract: Data about an accident may be obtained from a variety of sources including historical accident reports, geolocation data, weather data, and “hidden” data. These data may be combined to produce complete feature sets. Complete feature sets from a plurality of accidents may be used to generate model feature sets for driving behavior models.

Abstract: A path computing method, system, and computer program product, include extracting unpleasant data from a database to create a multi-variate spatio-temporal density function, collecting a tolerance level of a user, and computing a path for the user based on the tolerance level and the density function.