Handwriting Input Model Scheduling for Recognition Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional systems for handwriting recognition in languages with large character sets, such as Chinese, are impractical due to the time-consuming process of cycling through numerous character pages on keyboards, and simultaneous use of multiple language and character recognition models is computationally expensive.

Innovation Solution

A technique that schedules only one pair of language and character recognition models at a time on a client computing device, initially selecting default models for natural languages, and switches to alternative models like emoji models when initial probability scores do not meet a threshold, optimizing computational resources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple language and character recognition models are used simultaneously for handwriting recognition, then recognition accuracy is improved, but computational cost and processing time increase significantly

Engineering Contradiction:

Improverecognition accuracyVSAvoidcomputational cost

Core Design Contradiction:

Measurement precisionVSUse of energy by stationary object

Solution Approach 1:

The system dynamically switches between different model pairs based on the handwriting input characteristics. Instead of running multiple model pairs simultaneously, the scheduler selects and activates only the most appropriate model pair for the current input, reducing computational overhead while maintaining recognition accuracy across different scenarios

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the parameters of the recognition system by switching between different model configurations (different language models and character recognition models) based on the input characteristics. This allows the system to adapt to different handwriting styles and languages without the computational burden of running all models concurrently

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If multiple model pairs are activated to handle diverse handwriting inputs, then versatility is improved, but device performance and processing speed deteriorate

Engineering Contradiction:

Improvehandling diverse handwriting inputsVSAvoidprocessing speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The scheduler dynamically adapts the model configuration based on the characteristics of the handwriting input. By analyzing the input and selecting the most appropriate model pair, the system maintains high versatility for handling different languages and handwriting styles while avoiding the performance penalty of running multiple models simultaneously

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system performs preliminary analysis of the handwriting input to determine which model pair is most suitable before activating the recognition process. This preliminary action allows the system to prepare the optimal model configuration in advance, ensuring both versatility and processing efficiency

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If conventional keyboards are used for languages with large character sets, then character selection capability is improved, but user input time increases significantly

Engineering Contradiction:

Improvecharacter selection capabilityVSAvoiduser input time

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system replaces the mechanical keyboard selection process with an automated handwriting recognition system. Instead of requiring users to manually navigate through multiple pages of characters on a keyboard, the system uses machine learning models to automatically recognize and identify characters from handwriting inputs, dramatically reducing input time while maintaining comprehensive character selection capability

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS10325018B2Techniques for scheduling language models and character recognition models for handwriting inputs

Publication Date: 2019.06.18 GOOGLE LLC

AI summary

A first handwriting input is received comprising strokes corresponding to a set of first characters comprising one or more first characters forming a first language model unit. A set of candidate first characters and a set of candidate first language model units with corresponding probability scores are determined based on an analysis of the one or more sets of candidate first characters using the first language model and a corresponding first character recognition model. When no first probability score satisfies a threshold, one or more sets of candidate second characters and a set of candidate second language model units are determined based on an analysis of the first handwriting input using a second language model and a corresponding second character recognition model. A first candidate list is then output comprising at least one of the set of candidate second language model units.