Bounded Incremental Clustering via Subset Sampling

Solution Overview

Problem

Conventional data clustering techniques are time- and computationally expensive, especially when dealing with large datasets and high-dimensional data, as they often require re-clustering all existing data instances with new data, leading to unbounded compute time and resource consumption.

Innovation Solution

A clustering system that performs incremental clustering of new data instances with existing clusters in a bounded manner by selecting a subset of similar existing data clusters, sampling instances, forming intermediate clusters, and mapping them to existing clusters, thereby controlling compute time and resource consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional data clustering techniques are used to cluster new data instances with all existing data instances, then clustering accuracy is maintained, but compute time and resource consumption become unbounded

Engineering Contradiction:

Improveclustering accuracyVSAvoidcompute time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the existing data clusters into multiple subsets, where each subset contains a portion of the existing data instances. Instead of comparing new data instances with all existing instances, the system divides and conquers the clustering task by processing smaller segments, thereby reducing computational complexity while maintaining clustering quality through multiple passes or iterative refinement.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by selecting and processing only a representative subset of existing data instances rather than all instances. This subset may be chosen through sampling, filtering, or prioritization strategies that capture the essential characteristics of the full dataset, achieving sufficient clustering accuracy with reduced computational resources.

Inventive Principle:

Principle #16Partial or excessive action

2Measurement precision

If conventional data clustering techniques are used to cluster new data instances with all existing data instances, then clustering accuracy is maintained, but resource consumption increases

Engineering Contradiction:

Improveclustering accuracyVSAvoidresource consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the existing data clusters into multiple subsets, where each subset contains a portion of the existing data instances. Instead of comparing new data instances with all existing instances, the system divides and conquers the clustering task by processing smaller segments, thereby reducing computational complexity while maintaining clustering quality through multiple passes or iterative refinement.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by selecting and processing only a representative subset of existing data instances rather than all instances. This subset may be chosen through sampling, filtering, or prioritization strategies that capture the essential characteristics of the full dataset, achieving sufficient clustering accuracy with reduced computational resources.

Inventive Principle:

Principle #16Partial or excessive action

3Reliability

If conventional data clustering techniques are used to cluster new data instances with all existing data instances, then complete clustering is achieved, but memory requirements become unbounded

Engineering Contradiction:

Improveclustering completenessVSAvoidmemory requirements

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments the existing data clusters into multiple subsets, where each subset contains a portion of the existing data instances. Instead of comparing new data instances with all existing instances, the system divides and conquers the clustering task by processing smaller segments, thereby reducing computational complexity while maintaining clustering quality through multiple passes or iterative refinement.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by selecting and processing only a representative subset of existing data instances rather than all instances. This subset may be chosen through sampling, filtering, or prioritization strategies that capture the essential characteristics of the full dataset, achieving sufficient clustering accuracy with reduced computational resources.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20230385382A1Bounded incremental clustering

Publication Date: 2023.11.30 ADOBE INC

AI summary

A clustering system provides bounded incremental clustering for adding input data instances to existing data clusters. Input data instances are received and processed to form input data clusters. For a given input data cluster, a subset of existing data clusters is selected, and a subset of existing data instances are selected from each of the selected existing data clusters. The selected existing data instances and the input data instances from the given input data cluster are processed to form intermediate clusters. At least one intermediate cluster is mapped to an existing data cluster.