General Information Metrics for Improving AI Model Training Efficiency
This addresses the problem of reducing training costs for AI developers, but it appears incremental as it applies existing metrics to data selection.
The paper tackled the problem of high training costs due to large datasets and lack of universal data selection methods by proposing the GIME method, which reduced total model training expenses by 39.56% in a Judicial AI Program while preserving performance.
To address the growing size of AI model training data and the lack of a universal data selection methodology-factors that significantly drive up training costs -- this paper presents the General Information Metrics Evaluation (GIME) method. GIME leverages general information metrics from Objective Information Theory (OIT), including volume, delay, scope, granularity, variety, duration, sampling rate, aggregation, coverage, distortion, and mismatch to optimize dataset selection for training purposes. Comprehensive experiments conducted across diverse domains, such as CTR Prediction, Civil Case Prediction, and Weather Forecasting, demonstrate that GIME effectively preserves model performance while substantially reducing both training time and costs. Additionally, applying GIME within the Judicial AI Program led to a remarkable 39.56% reduction in total model training expenses, underscoring its potential to support efficient and sustainable AI development.