AT&T Intellectual Property I, LP

Salesforce, Inc.

Donglin Hu

Jin Wang

Zhengye Liu

Yali Liu

Bo Han

Kexin Xie

Yuxi Zhang

Chen Xu

A prediction system of an online system deploys one or more machine-learned architectures to generate predictions. In one embodiment, the machine-learned architecture is a stacked ensemble model. The stacked ensemble model includes a plurality of base models, where a base model is coupled to receive input data and generate a base prediction for the input data. The stacked ensemble model includes a meta model that combines the base predictions to generate a meta prediction for the input data. The prediction system also generates a reliability measure that takes advantage of the base predictions to evaluate the reliability of the meta prediction. In this manner, while the quality of individual predictions may differ from one another depending on the values of the input data, the prediction system can dynamically generate the reliability measure to account for this variation.

Generating reliability measures for machine-learned architecture predictions

A method includes obtaining a first plurality of encrypted traffic flows traversing a communication network, performing a first classification, wherein a result of the first classification identifies a traffic type associated with each encrypted traffic flow of the first plurality of encrypted traffic flows, and wherein the first classification is based on a traffic pattern of the each encrypted traffic flow, performing a second classification, wherein a result of the second classification identifies a traffic type associated with each server name indication from which the first plurality of encrypted traffic flows is associated, and wherein the second classification is based on the result of the first classification, and performing a third classification identifying a traffic type associated with each encrypted traffic flow of the first plurality of encrypted traffic flows, wherein the third classification is based on a combination of the results of the first classification and the second classification.

Classification of encrypted internet traffic

Aspects of the subject disclosure may include, for example, receiving a first plurality of operating parameters associated with a first plurality of communication devices, obtaining a first prediction of a performance of a first communication device included in the first plurality of communication devices that is based on an application of the first plurality of operating parameters to a model, and responsive to a determination that the first prediction of the performance is inadequate in accordance with a first threshold, transmitting a first command to at least one communication device included in the first plurality of communication devices to modify at least one operating parameter of the at least one communication device. Other embodiments are disclosed.

Method and apparatus for utilizing radio access network guidance to select operating parameters

Aspects of the subject disclosure may include, for example, a method, including identifying recommended video clips for a user of a communication device according to a video viewing profile for the user and video subject matter information associated with a plurality of video clips, identifying an edge cloud server for facilitating network access by the communication device at a location, determining a reduced activity period for a data path between a video content server and the edge cloud server by comparing activity information for the data path and an activity threshold, directing the video content server to store the recommended video clips at the edge cloud server during the reduced activity period, and providing a listing of the recommended video clips to an application, where the communication device receives a video clip of the recommended video clips from the edge cloud server responsive to a selection of the video clip via the application. Other embodiments are disclosed.

Method and apparatus for delivery of media content

Aspects of the subject disclosure may include, for example, applying a model to a first unlabeled traffic item to generate a classification of the first unlabeled traffic item as a first class of traffic items or a second class of traffic items, generating a first score that is representative of a first confidence of the classification, determining that the first score is greater than a first threshold, and responsive to determining that the first score is greater than the first threshold, refining the model in accordance with at least the classification, resulting in a modified model. Other embodiments are disclosed.

Apparatus and method to facilitate an iterative, machine learning based traffic classification

Concepts and technologies disclosed herein are directed to collective intelligence-based cell congestion detection in mobile telecommunications networks. According to some aspects of the concepts and technologies disclosed herein, a core network element or device application can detect cell congestion based upon Internet layer, transport layer, and application layer data, such as, for example, traffic type, volume, rate, latency, jitter, and the like. According to one aspect disclosed herein, a cell congestion detection ('CCD') system can collect data from an Internet layer, a transport layer, and an application layer. The CCD system can apply a machine learning algorithm to the data to determine whether the cell is congested. According to another aspect the CCD system can collect the data that is associated with a specific application executed by a plurality of UE devices connected to a cell.

Collective intelligence-based cell congestion detection in mobile telecommunications networks

Aspects of the subject disclosure may include, for example, a device in which a processing system enables mobile devices on a communication network to perform latency tests to obtain first latency test data, determines an edge latency for each mobile device resulting in edge latencies, identifies a group of mobile device clusters according to the edge latencies, and identifies network edge server locations according to the group of mobile device clusters. The processing system also receives second latency test data from each of the network edge servers, and determines a data center latency for each of the network edge servers, resulting in a plurality of data center latencies. The processing system also identifies a data center location according to the data center latencies, and determines a network topology for the communication network; the network topology includes the data center location and the edge server locations. Other embodiments are disclosed.

Growing community of inventors

Dublin, CA, United States of America

Donglin Hu