Abstract
Techniques are described herein for using semi-supervised machine learning to simplify an intent interface for end users by allowing a user to specify key network features in which they are interested. A continual learning based approach better adapts to a continuously changing intent interface and simplifies the experience for end users. The semi-supervised learning algorithm learns the reverse mapping (stored in an intent cache database) of Group-Based Policy (GBP) policy templates expressed using data models (e.g., as Yang models) as well as user network feature key words given a set of existing network configuration use cases provided as topology network maps, device configurations, and manually crafted GBP policy objects. A new user starts by specifying key intent features of interest, picks the closest mapping GBP template, and configures their network.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Rustogi, Sudhir, "SEMI-SUPERVISED MACHINE LEARNING OF INTENT DATA MODELS BASED ON GROUP BASED POLICY", Technical Disclosure Commons, (October 31, 2018)
https://www.tdcommons.org/dpubs_series/1622