Efficient Compressive Strength Prediction of Alkali-Activated Waste Materials Using Machine Learning.

This study explores the integration of machine learning (ML) techniques to predict and optimize the compressive strength of alkali-activated materials (AAMs) sourced from four industrial waste streams: blast furnace slag, fly ash, reducing slag, and waste glass. Aimed at mitigating the labor-intensive trial-and-error method in AAM formulation, ML models can predict the compressive strength and then streamline the mixture compositions. By leveraging a dataset of only 42 samples, the Random Forest (RF) model underwent fivefold cross-validation to ensure reliability. Despite challenges posed by the limited datasets, meticulous data processing steps facilitated the identification of pivotal features that influence compressive strength. Substantial enhancement in predicting compressive strength was achieved with the RF model, improving the model accuracy from 0.05 to 0.62. Experimental validation further confirmed the ML model's efficacy, as the formulations ultimately achieved the desired strength threshold, with a significant 59.65% improvement over the initial experiments. Additionally, the fact that the recommended formulations using ML methods only required about 5 min underscores the transformative potential of ML in reshaping AAM design paradigms and expediting the development process.

Exploring Dielectric Constant and Dissipation Factor of LTCC Using Machine Learning.

Low-temperature co-fired ceramics (LTCCs) have been attracting attention due to rapid advances in wireless telecommunications. Low-dielectric-constant (Dk) and low-dissipation-factor (Df) LTCCs enable a low propagation delay and high signal quality. However, the wide ranges of glass, ceramic filler compositions, and processing features in fabricating LTCC make property modulating difficult via experimental trial-and-error approaches. In this study, we explored Dk and Df values of LTCCs using a machine learning method with a Gaussian kernel ridge regression model. A principal component analysis and k-means methods were initially performed to visually analyze data clustering and to reduce the dimension complexity. Model assessments, by using a five-fold cross-validation, residual analysis, and randomized test, suggest that the proposed Dk and Df models had some predictive ability, that the model selection was appropriate, and that the fittings were not just numerical due to a rather small data set. A cross-plot analysis and property contour plot were performed for the purpose of exploring potential LTCCs for real applications with Dk and Df values less than 10 and 2 × 10-3, respectively, at an operating frequency of 1 GHz. The proposed machine learning models can potentially be utilized to accelerate the design of technology-related LTCC systems.