Dynamic Pruning Threshold Curved Surface for Hypothesis Search

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data processing devices for hypothesis search often incorrectly prune hypotheses due to static or dynamically adjusted pruning thresholds, leading to search errors, especially when multiple pruning measures are used and one threshold is exceeded, resulting in reduced recognition accuracy and speed.

Innovation Solution

A data processing device that plots hypotheses on a threshold space using multiple pruning measures, sets isopycnic surfaces based on hypothesis densities, generates a threshold curved surface where a decrease in one measure causes an increase in another, and uses the intersection of this surface with a hypothesis curved surface to determine pruning thresholds, allowing for dynamic adjustment of pruning measures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a fixed or dynamically adjusted pruning threshold is used, then the search efficiency is improved, but the recognition accuracy deteriorates due to incorrect pruning of hypotheses

Engineering Contradiction:

Improvesearch efficiencyVSAvoidrecognition accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent applies dynamics by transitioning from fixed or simply dynamically adjusted thresholds to a threshold that adapts based on the real-time distribution of hypothesis scores. The threshold curved surface is dynamically updated according to the current hypothesis set, allowing the pruning threshold to flexibly adjust to different search conditions and maintain both efficiency and accuracy.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of the pruning threshold from a static or linearly adjusted value to a curved surface parameter that reflects the actual score distribution of hypotheses. By modeling the threshold as a curved surface in the score difference space, the system can accurately capture the non-linear relationships between score differences and pruning decisions, thereby improving both efficiency and accuracy.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If multiple pruning measures are used with independent thresholds, then the coverage of pruning conditions is improved, but the reliability deteriorates due to incorrect pruning when one measure exceeds its threshold

Engineering Contradiction:

Improvepruning condition coverageVSAvoidpruning accuracy

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent merges multiple independent pruning measures (score difference threshold and hypothesis number threshold) into a unified threshold curved surface model. This integration allows the system to consider the combined effects of multiple pruning measures simultaneously, ensuring that hypotheses are only pruned when they fail to meet the comprehensive criteria defined by the curved surface, thereby improving reliability while maintaining adaptability.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a composite pruning threshold model that combines different pruning criteria (score difference and hypothesis number) into a single integrated structure. The threshold curved surface acts as a composite material that incorporates the properties of multiple pruning measures, allowing the system to leverage the strengths of each measure while mitigating their individual weaknesses through synergistic interaction.

Inventive Principle:

Principle #40Composite materials

3Reliability

If the pruning threshold is set to be lenient, then the recognition accuracy is improved by retaining more hypotheses, but the productivity deteriorates due to increased calculation amount

Engineering Contradiction:

Improverecognition accuracyVSAvoidrecognition speed

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent applies local quality by making the pruning threshold locally adaptive rather than uniformly applied. The threshold curved surface adjusts the pruning strictness based on the local characteristics of the hypothesis score distribution in different regions of the search space. This allows the system to be more lenient in regions where hypotheses are dense (preserving accuracy) and more strict in regions where hypotheses are sparse (improving speed), thereby optimizing the trade-off between accuracy and productivity.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS9047562B2Data processing device, information storage medium storing computer program therefor and data processing method

Publication Date: 2015.06.02 NEC CORP

AI summary

A plurality of pruning measures (PM) are calculated from a feature amount (CV) of test data (TD) which is input, a plurality of isopycnic surfaces (EC) are plotted and set on a threshold space (SS), a threshold curved surface (SC) in which a decrease in at least one of a plurality of pruning measures (PM) causes an increase in at least one thereof is generated using a portion of one isopycnic surface (EC) as a part, a hypothesis curved surface (HC) of subject data (CD) is generated on the threshold space (SS) to set a position intersecting the threshold curved surface (SC) to a pruning threshold (PS), and a plurality of hypotheses of the subject data (CD) are pruned. Thereby, there is provided a data processing device of which at least one of the recognition speed and the recognition accuracy is higher than in the related art.