Learning Control System for Intent Estimation Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
General users without programming expertise face difficulties in creating customized AI-based bots, as existing methods require specialized knowledge and are not user-friendly for constructing machine learning models to estimate input utterances effectively.
Innovation Solution
A method and apparatus that allow non-programming specialists to control the learning of machine learning models for estimating input utterances by providing visualized indices for decision-making, enabling the selection of learning targets and training based on registered utterances, using a processor to manage the learning process and interface for inputting and evaluating utterances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a machine learning model is constructed using traditional programming methods, then the model can estimate input utterances with high accuracy, but general users without programming expertise cannot easily create customized bots
Solution Approach 1:
The patent introduces an intermediary platform that mediates between the user and the machine learning model construction process. This platform provides a graphical user interface with visualized indices (first index for utterance registration status, second index for learning level) that translates complex model training parameters into user-friendly visual indicators, enabling general users to construct customized bots without programming expertise while maintaining model accuracy
Solution Approach 2:
The patent transforms the complex parameters of machine learning model construction into simplified visual indices. The first index displays the registration status of utterances for each intention, and the second index shows the learning level, allowing users to make informed decisions about which intentions to train based on visual feedback rather than understanding complex technical parameters
2Reliability
If comprehensive learning data is collected for all intentions, then the model estimation accuracy improves, but the time and resources required for training increase significantly
Solution Approach 1:
The patent enables users to perform partial learning actions by selecting specific intentions as learning targets based on visualized indices. Users can choose to train only the intentions that require improvement (identified through the first and second indices) rather than retraining all intentions, thereby reducing training time and resources while maintaining sufficient model accuracy for the most critical functions
Solution Approach 2:
The patent implements a feedback mechanism where the system provides visualized indices showing the current state of utterance registration and learning levels for each intention. This feedback allows users to make informed decisions about which intentions need additional training data or retraining, optimizing the allocation of training resources to achieve the best accuracy improvement with minimal time investment
Data Source
AI summary
A method and apparatus for controlling learning of a model for estimating an intention of an input utterance is disclosed. A method of controlling learning of a model for estimating an intention of an input utterance among a plurality of intentions includes providing a first index corresponding to the number of registered utterances for each intention, providing a second index corresponding to a learning level for each intention, providing a learning target setting interface such that at least one intention that is to be a learning target is selected from among the intentions based on the first index and the second index, and training the model based on the registered utterances for each intention and setting of the learning target for each intention.