Adaptive ML Binning for EDA Resource Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic design automation (EDA) tools face challenges in resource constraints, such as dynamic changes in computational resources, license fees, and time-to-results, which affect the accuracy and efficiency of machine learning classification models, especially when dealing with small, noisy, and imbalanced training data.
Innovation Solution
The implementation of an adaptive classification and decision system that dynamically determines bin thresholds and selects processes based on costs and a global budget, optimizing the classification of discrete probabilities to reduce resource consumption and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning classification models are applied to EDA tools, then classification accuracy can be improved, but resource consumption increases
Solution Approach 1:
The patent segments the classification process by dividing discrete probabilities into multiple bins based on slope changes. This segmentation allows the system to process and analyze probability distributions in manageable segments, improving classification accuracy while controlling computational resource consumption through structured data organization.
Solution Approach 2:
The patent applies partial action by selectively evaluating bins based on a global budget constraint. The system evaluates bins consecutively and stops when the budget is exhausted, performing only the necessary portion of the classification task required to achieve acceptable results within resource limits, rather than exhaustively processing all data.
2Reliability
If more EDA tools are used for testing, then defect detection accuracy improves, but cost and time increase
Solution Approach 1:
The patent implements partial action by evaluating bins consecutively against a global budget and stopping when resources are exhausted. This allows the system to achieve acceptable defect detection accuracy using only a portion of available EDA tools and time, rather than exhaustively testing with all tools, thus reducing time-to-results while maintaining reliability within constraints.
3Productivity
If computational resources are increased, then machine learning model performance improves, but license fees and operational costs increase
Solution Approach 1:
The patent applies dynamics by making the bin evaluation process adaptive to available resources. The system dynamically evaluates bins consecutively based on a global budget constraint, adjusting the extent of processing according to available computational resources and costs, thereby optimizing model performance within varying resource conditions rather than requiring fixed high resource allocation.
Solution Approach 2:
The patent uses partial action by selectively processing bins until the global budget is exhausted. This allows the system to achieve acceptable machine learning model performance using only a portion of available computational resources, reducing license fees and operational costs while maintaining productivity within budgetary constraints.
Data Source
AI summary
Optimizing ML models for resource constraints in electronic design automation (EDA) computer aided design (CAD) flows, including computing a set of bin thresholds based on slope changes in an ordered set of discrete probabilistic classification scores, assigning the discrete probabilistic classification scores to the bins based on the values of the discrete probabilistic classification scores and the bin thresholds, and selecting processes associated with the discrete probabilistic classification scores of one or more of the bins based on costs of the respective processes and a global budget.