Computer Analysis of Trained Machine Learning Models
The described method and system generate a database to analyze machine learning models, addressing the challenge of understanding their training and behavior, thereby improving model performance and accuracy through automated evaluation and retraining recommendations.
JP2025524174APending Publication Date: 2025-07-25CITRUSX LTD
View PDF 0 Cites 0 Cited by
Patent Information
- Application Number
- JP2025504632
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-07-27
- Filing Date
- 2023-07-24
- Publication Date
- 2025-07-25
AI Technical Summary
Technical Problem
Existing machine learning models are cumbersome to understand whether they are properly trained and difficult to analyze their behavior, making it challenging to improve their training and predict their performance accurately.
Method used
A method and system using a processor and memory circuit to generate a database with data points that provide information on input and prediction vectors, allowing for the analysis of machine learning models, including determining their quality, confidence, and recommending retraining or input vector modifications.
Benefits of technology
Enables automatic understanding of a machine learning model's training quality, confidence in predictions, and identifies areas for improvement, facilitating more efficient and accurate model performance.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure 2025524174000001_ABST
Abstract
A system and method, by a processor and a memory circuit, for a machine learning model, to obtain a set of data points providing information of a set of input vectors, wherein each input vector in the set of input vectors is one used to train the machine learning model, and a set of prediction vectors, to use the set of data points to generate a database providing information of the machine learning model, the database providing information of terminal nodes, each given terminal being associated with one or more coefficients defining a function that fits, with a quality meeting an accuracy criterion, the relationship between a plurality of data points of the set, a plurality of input vectors of the given terminal node, and a plurality of prediction vectors of the given terminal node, the system and method being provided, wherein the database is usable to generate data providing information of the machine learning model.
Need to check novelty before this filing date? Find Prior Art