17809021. DYNAMICALLY FEDERATED DATA BREACH DETECTION simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)

From WikiPatents
Jump to navigation Jump to search

DYNAMICALLY FEDERATED DATA BREACH DETECTION

Organization Name

INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor(s)

Divyesh Jadav of San Jose CA (US)

Mu Qiao of Belmont CA (US)

Eric Kevin Butler of San Jose CA (US)

DYNAMICALLY FEDERATED DATA BREACH DETECTION - A simplified explanation of the abstract

This abstract first appeared for US patent application 17809021 titled 'DYNAMICALLY FEDERATED DATA BREACH DETECTION

Simplified Explanation

The abstract describes a system where a trained machine learning model and feature information are distributed from a server to multiple client devices. Each client device trains its own machine learning model using local data and constructs an unsupervised model using the feature information. The system then determines when there is a significant difference in performance between the supervised and unsupervised models and proposes changes to the feature information. If the proposed change improves the performance of a client device, it is communicated to a sample set of devices. If the majority of the sampled devices show improved performance, the change is communicated back to the server.

  • A trained machine learning model and feature information are distributed from a server to multiple client devices.
  • Each client device trains its own machine learning model using local data.
  • Each client device constructs an unsupervised model using the feature information.
  • The system determines when there is a significant performance difference between the supervised and unsupervised models.
  • Proposed changes to the feature information are identified.
  • The proposed change is deployed on one client device.
  • If the proposed change improves the performance of that client device, it is communicated to a sample set of devices.
  • If the majority of the sampled devices show improved performance, the change is communicated back to the server.

Potential Applications

This technology can be applied in various domains where distributed machine learning models are used, such as:

  • Fraud detection systems
  • Anomaly detection in network security
  • Predictive maintenance in industrial settings
  • Personalized recommendation systems

Problems Solved

This technology addresses the following problems:

  • Centralized training of machine learning models may not be feasible or efficient for large-scale deployments.
  • Local data on client devices may have unique characteristics that can improve the performance of the models.
  • The detection performance of supervised and unsupervised models may differ significantly, requiring adjustments to the feature information.

Benefits

The benefits of this technology include:

  • Improved detection performance by leveraging local data and unsupervised models.
  • Efficient distribution of trained models and feature information to client devices.
  • Adaptive system that can identify and communicate changes that improve performance.
  • Reduced reliance on centralized server for model training and updates.


Original Abstract Submitted

A processor distributes, from a server, a trained supervised machine learning (ML) model and supervised and unsupervised feature information to a plurality of client devices; at each client device, trains the supervised ML model using local data to generate a local supervised ML model, constructs a local unsupervised ML model using the unsupervised feature information, and deploys the local supervised and unsupervised ML models; determining when a detection performance difference between the local supervised and unsupervised ML models reaches a threshold; identifies a proposed change to the supervised or unsupervised feature information; deploys the proposed change on one client device; responsive to determining the proposed change improves the detection performance of that client device, communicates the proposed change to a sampled set of client devices; and responsive to determining the proposed change improves the detection performance of a majority of the sampled set, communicates the proposed change to the server.

Retrieved from "http://wikipatents.org/index.php?title=17809021._DYNAMICALLY_FEDERATED_DATA_BREACH_DETECTION_simplified_abstract_(INTERNATIONAL_BUSINESS_MACHINES_CORPORATION)&oldid=17768"