Wafer Fault detection in manufacturing using Machine Learning

In semiconductor manufacturing, the imperative of producing defect-free wafers is paramount. This project focuses on wafer fault detection, utilizing a dataset comprising both wafer maps and tabular data. Notably, this employes a Convolutional Autoencoder (CAE) model for data augmentation, enriching the dataset and fortifying the robustness of defect detection models. The objectives encompass the development of AI models for precise defect prediction from tabular data, coupled with the CAE model’s application to augment wafer maps. This integrated strategy automates defect detection and facilitates visual representation on wafer maps, thereby enhancing manufacturing efficiency, minimizing defects, and elevating product quality. The convergence of advanced AI techniques and data augmentation embodies a pioneering approach in semiconductor manufacturing. By automating defect detection and refining visual representation, this project heralds a transformative era, improving manufacturing efficiency, and significantly reducing defects. It elevates efficiency, reduces errors and fosters sustainability. Beyond technology, it represents a convergence of human ingenuity and $\mathbf{A I}$, redefining production with precision, reliability and a new standard of excellence.