Bayesian Classifier Using Non-Linear Probability Functions

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Classification systems face challenges in accurately distinguishing between objects of interest and nuisances using Boolean attributes and linear probability functions, particularly in complex scenarios like differentiating between oncoming vehicle headlights and leading vehicle taillights based on non-Boolean attributes like brightness and color.

Innovation Solution

A Bayesian classifier system employing non-linear probability functions processes non-Boolean attributes, such as brightness and color, to determine the likelihood and probability of objects being classified as either an object of interest or a nuisance, using a processor connected to an imager that captures forward scenes, enabling accurate classification by combining likelihood functions and weighting attributes appropriately.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If Boolean attributes and linear probability functions are used for classification, then the system is simple to implement, but the classification accuracy is insufficient for complex scenarios

Engineering Contradiction:
Improveclassification accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent transforms the classification approach by changing the parameter type from Boolean (binary) attributes to continuous non-Boolean attributes, and from linear probability functions to non-linear probability density functions. This allows the system to capture complex relationships between attributes and class labels, significantly improving classification accuracy for distinguishing headlights from taillights while managing complexity through efficient computational methods.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If non-Boolean attributes are used to classify objects, then the classification accuracy improves, but the computational complexity increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidcomputational power
Core Design Contradiction:
Measurement precisionVSPower

Solution Approach 1:

The patent replaces complex iterative computational mechanisms with a direct mathematical approach using non-linear probability density functions. By formulating the classification problem in terms of likelihood ratios and probability densities, the system achieves high accuracy with reduced computational overhead, avoiding the need for extensive iterative optimization while still capturing complex attribute relationships.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If linear probability functions are used, then the computation is fast, but the ability to distinguish between similar objects like headlights and taillights is limited

Engineering Contradiction:
Improveobject differentiation accuracyVSAvoidcomputation time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent introduces non-linearity into the probability functions, analogous to replacing straight lines with curves. By using non-linear probability density functions that can model complex decision boundaries, the system achieves superior object differentiation between headlights and taillights. The non-linear functions capture curved decision surfaces in the feature space, enabling accurate classification while maintaining computational efficiency through closed-form solutions.

Inventive Principle:
Principle #14Spheroidality (Curvature)

Data Source

PatentEP2801057B1Bayesian classifier system and method thereof
Publication Date: 2019.04.10 GENTEX CORP
  • EP2801057B1 patent drawingFigure 1~1A
  • EP2801057B1 patent drawingFigure 2
  • EP2801057B1 patent drawingFigure 3

AI summary

A classification system and method are provided, wherein the classification system includes a memory device, a processor communicatively connected to the memory device, and an input communicatively connected to the processor, wherein the input is configured to receive data comprising at least one object that is to be classified as one of an object of interest (OOI) and a nuisance of interest (NOI) based upon at least one non-Boolean attribute of the object, wherein the processor is configured as a Bayesian classifier to classify the object based upon the non-Boolean attribute using a non-linear probability function.