Adaptive Online Data Protection System Using Machine Learning Feedback

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems for controlling unauthorized data activity rely on hard-coded rules, leading to numerous false positives and false negatives, and lack the flexibility to learn from user actions, making them ineffective in differentiating between authorized and unauthorized data access.

Innovation Solution

An adaptive online data protection system using machine learning techniques to identify and respond to anomalous data activity, allowing for real-time differentiation between authorized and unauthorized data access without the need for pre-defined rules, and learns from user feedback to refine its decisions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If hard-coded rules are used to control data activity, then data security policies can be enforced, but the system generates numerous false positives and false negatives and cannot learn from user actions

Engineering Contradiction:
Improvedata security enforcementVSAvoidability to learn from user actions
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system implements feedback loops where user responses to alerts (false positives and false negatives) are collected and used to retrain the machine learning model. This continuous feedback mechanism allows the system to learn from its mistakes and improve its accuracy over time, directly addressing the inability of hard-coded rules to adapt to user behavior patterns

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The machine learning model performs self-improvement by automatically learning from user feedback without requiring manual rule updates. The system serves itself by continuously refining its own detection algorithms based on accumulated data, transitioning from static hard-coded rules to a dynamic self-learning system

Inventive Principle:
Principle #25Self-service

2Reliability

If hard-coded rules are used for data activity control, then unauthorized access can be detected, but the system produces many false positives and false negatives

Engineering Contradiction:
Improveunauthorized access detectionVSAvoiddetection accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The system changes the fundamental parameter of detection from static rule-based thresholds to dynamic machine learning predictions. The ML model analyzes multiple features and patterns simultaneously, adjusting detection parameters based on learned relationships, which significantly improves detection precision and reduces false positives compared to rigid hard-coded rules

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The detection system combines multiple data sources and analysis methods into a composite approach. Instead of relying on a single hard-coded rule, the system integrates various user behavior patterns, access contexts, and historical data through machine learning, creating a more robust and precise detection mechanism

Inventive Principle:
Principle #40Composite materials

3Measurement precision

If machine learning techniques are used to identify anomalous data activity, then false positives and false negatives are reduced, but the system requires continuous learning from user feedback

Engineering Contradiction:
Improvedetection accuracyVSAvoidsystem learning mechanism
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system implements automated feedback collection and processing mechanisms where user responses to alerts are automatically captured, processed, and used to retrain the model. This automated feedback loop manages the complexity by systematizing the learning process rather than requiring manual intervention, making the continuous learning aspect manageable

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent introduces an intermediary alerting system that acts as a buffer between the ML detection model and the user. The system presents uncertain cases as alerts for user review, allowing the model to learn from user decisions without directly exposing the complexity of the learning mechanism to end users

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3679504B1Adaptive online data activity protection
Publication Date: 2024.07.31 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP3679504B1 patent drawingFigure 1
  • EP3679504B1 patent drawingFigure 2~3
  • EP3679504B1 patent drawingFigure 4

AI summary

Systems and methods for identifying and responding to anomalous data activity by a computer user on a computing device are presented. An anomalous data activity service, implemented as a machine learning service, receives notice of data activity and conducts an evaluation to determine whether the data activity is an anomalous data activity. Upon determining that the data activity is an anomalous data activity, a responsive action may be taken that may result in the anomalous data activity being blocked or allowed.