17524020. MACHINE LEARNING MODEL CHANGE DETECTION AND VERSIONING simplified abstract (International Business Machines Corporation)

From WikiPatents
Jump to navigation Jump to search

MACHINE LEARNING MODEL CHANGE DETECTION AND VERSIONING

Organization Name

International Business Machines Corporation

Inventor(s)

Shubhi Asthana of Santa Clara CA (US)

Shikhar Kwatra of San Jose CA (US)

Sushain Pandit of Austin TX (US)

MACHINE LEARNING MODEL CHANGE DETECTION AND VERSIONING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17524020 titled 'MACHINE LEARNING MODEL CHANGE DETECTION AND VERSIONING

Simplified Explanation

Systems, methods, and computer programming products for versioning machine learning models are disclosed in this patent application. The technology detects, quantifies, and compares changes between new and existing datasets using statistical and semantic feature comparisons. The patent application provides recommendations for versioning existing models based on these changes in feature importance. Here are the key points:

  • The technology detects changes between datasets used in machine learning models and new datasets that introduce new features or evolving features.
  • Statistical and semantic feature comparisons are used to quantify and compare these changes.
  • Recommendations are provided based on the changes in feature importance, statistical changes, and semantic feature comparisons.
  • The recommendations describe whether models should be updated with a re-trained model or if the existing features do not indicate a need for re-training.

Potential applications of this technology:

  • Machine learning model versioning and updating
  • Data analysis and comparison in machine learning applications
  • Recommendation systems for model updates based on dataset changes

Problems solved by this technology:

  • Ensures that machine learning models are up-to-date with evolving datasets
  • Provides a systematic approach to detect and quantify changes in datasets
  • Helps in making informed decisions on whether to update machine learning models or not

Benefits of this technology:

  • Improves the accuracy and performance of machine learning models by incorporating new and relevant features
  • Saves time and resources by avoiding unnecessary re-training of models when the existing features are still relevant
  • Provides a systematic and data-driven approach to model versioning and updating.


Original Abstract Submitted

Systems, methods, and computer programming products for versioning machine learning models. Changes between new and existing datasets are detected, quantified and compared using statistical and semantic feature comparisons. Recommendations for versioning existing models are in response to detecting changes between the feature importance of datasets used in the application of the machine learning model and new datasets that introduce new features or features that evolve over time in such a manner that feature importance has shifted away from one or more features of the first dataset to the new dataset. Based on the changes in feature importance, statistical changes and semantic feature comparisons, the recommendations provided describe whether models should be updated with a re-trained model, or that the existing features of the model do not indicate a need for re-training.

Retrieved from "http://wikipatents.org/index.php?title=17524020._MACHINE_LEARNING_MODEL_CHANGE_DETECTION_AND_VERSIONING_simplified_abstract_(International_Business_Machines_Corporation)&oldid=22376"