Supporting Information

This supplementary document accompanies the article “Data-Driven Design of Dielectric Stability in BaTiO₃–Bi(Mg₁/₂Ti₁/₂)O₃ Ceramics Using Automated Machine Learning.” The BaTiO₃–Bi(Mg₁/₂Ti₁/₂)O₃ (BT–BMT) ceramic system has attracted considerable attention for multilayer ceramic capacitors (MLCCs) due to its promising dielectric stability at elevated temperatures. In this work, a data-driven automated machine learning (AutoML) framework was employed to accelerate the exploration of processing–composition–property relationships within BT–BMT ceramics. By integrating Auto-Sklearn and Auto-Keras algorithms, robust models were established to identify processing conditions that fulfill the X9R temperature stability standard (−55 to 200 °C, capacitance variation within ±15%). The Auto-Sklearn model exhibited excellent predictive accuracy (R² = 0.9929), enabling efficient screening of unexplored parameter spaces. Key factors influencing dielectric properties were identified, including sintering temperature, coarse grinding time, post-calcination grinding time, and sintering time. The optimized parameter combinations identified by the models were in line with experimental trends reported in literature, suggesting the reliability of the approach. These results demonstrate that combining data-driven approaches with ceramic process optimization provides an effective route to accelerate the design of high-performance dielectric ceramics.