In this work, we describe the development and implementation of a system for the automatic classification of PSG electronic circuit boards using machine learning methods. We provide an overview of the entire process, from data preprocessing to integrating the solution into the production process. As we are dealing with highly imbalanced data with two classes for the target variable, we describe options for manipulating the training dataset through the principle of oversampling the minority class.
We provide a more detailed description of the machine learning algorithms used and the TPE algorithm, which is based on the Bayesian approach to finding optimal hyperparameter values. Due to the specific nature of the data and the problem itself, we must define a special weighted metric for finding the optimal configuration and evaluating the model.
We then give a detailed description of the implementation and the model prediction process, followed by the integration method into the production line. We conclude with the presentation of the results.
|
