Enhanced Nanoparticle Classification Through Optimized Artificial Neural Networks

This study investigates the effectiveness of Deep Learning techniques in classifying nanoparticles, with a par-ticular focus on how the depth of neural networks influences model performance. Artificial Neural Networks (ANN) were utilized in simulations, exploring configurations ranging from a single hidden layer to four hidden layers to identify the optimal architecture for nanoparticle classification using MATLAB- NN Toolbox. Evaluation metrics include Mean Squared Error (MSE), Root Mean Squared Error (RMSE), the Correlation Coefficient (R), and the Coefficient of Determination <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(R^{2})$</tex>. Notably, the minimum value of the MSE achieved was <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$5.6814 \times 10^{-13}$</tex> with the Bayesian Regularization (BR) technique, while the maximum values of Rand <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> were 0.9989 and 0.9977, respectively, with the Levenberg-Marquardt (LM) technique. The results offer insights into model robustness and contribute to selecting an optimal algorithm for analyzing polymeric nanoparticles. This study highlights the potential of ANN to advance precision in nanomaterial classification tasks, supporting further development in this field.