Gaussian Process Parameter Setting for Faster Accurate Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in automatically and efficiently setting adjustable parameters for devices in industrial facilities and business transactions, requiring domain knowledge and manual decision-making, which can be slow and prone to outdated rules.
Innovation Solution
A computer-implemented recommendation method using a Gaussian process-based recommender system that processes input variables to determine optimal settings for adjustable parameters, incorporating a separable kernel for similarity measurement and collaborative filtering with matrix decomposition, enabling data-driven and self-optimizing decision support.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual decision-making based on domain knowledge and rule-based guidelines is used to set adjustable parameters, then the settings can be made with expert judgment, but the process becomes slow and the rules quickly become outdated in dynamic environments
Solution Approach 1:
The system enables self-service by allowing the recommender system to automatically learn from historical data and continuously optimize parameter settings without requiring manual updates of rules. The system serves itself by automatically adapting to new patterns in the data, eliminating the need for periodic manual rule updates while maintaining high accuracy in setting recommendations.
Solution Approach 2:
The patent replaces the mechanical system of manual rule-based decision-making with an automated machine learning-based recommender system. Instead of relying on static rules that require manual updates, the system uses trained models that automatically process input variables and generate setting recommendations, dramatically reducing decision time while maintaining or improving accuracy.
2Loss of information
If more data is collected from customers and transactions to improve decision quality, then better insights can be gained, but the complexity of extracting and evaluating the data increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the complex data processing task into distinct modules: data collection from multiple sources, data preprocessing and cleaning, feature extraction, model training, and recommendation generation. This modular approach allows the system to handle large volumes of diverse data without overwhelming complexity, as each module processes a specific aspect of the data independently.
Solution Approach 2:
The recommender system acts as an intermediary between raw data and decision-making processes. It includes intermediate processing layers such as data preprocessing pipelines, feature engineering modules, and model training components that transform raw data into actionable insights. This intermediary structure manages complexity by providing structured transformation steps between data collection and final recommendations.
3Productivity
If automatic recommendation systems are implemented to speed up parameter setting, then decision speed increases, but the system requires sophisticated modeling functions and training data
Solution Approach 1:
The patent applies preliminary action by implementing an offline training phase where the recommender system is trained on historical data before deployment. The modeling functions are pre-trained and stored, so that during actual operation, the system only needs to query the pre-trained models with current input variables to generate rapid recommendations. This separates the complex learning process from the fast inference process, achieving both speed and sophistication.
Data Source
AI summary
Various embodiments of the teachings herein include a method for providing a setting for a given parameter to be adjusted. The method may include: providing a plurality n with n≥2 of given input variables VARv with v=1, . . . , n to a prepared recommender system; processing the input with a modeling function to determine the recommended setting; and setting the given parameter to the recommended setting. Input variables VARv correspond to different variable types VARTv. For a particular variable type VARTv a plurality Tv of respective variables VARTv,t is available with t=1, . . . , Tv. For each variable type VARTv only one variable VARTv,t is provided as input variable VARv=VARTv,t. The modeling function is a function trained based on a Gaussian process with DCOM˜(0,Kcoreg+σ2I) defined by a characterizing covariance matrix Kcoreg and a corresponding characterizing kernel kSEPcoreg.