Artificial intelligence techniques for material science application

Paper ID : 1102-ISCHU

Authors

Asmaa Abdelbaky * Cairo

Abstract

Researchers in academia and industry have shown interest in imaging techniques used for the characterization of nanomaterials. The literature presents the characterization of nanomaterials with properties tailored to meet the requirements of various applications. However, traditional methods are limited due to the increase in image size and complexity of content. Recently, advances in machine learning have been employed to enhance computers’ ability to comprehend these images . Materials Science is increasingly handling artificial intelligence methods to address the complexity in the field of everyday life necessities. This research will study the effect of applying deep learning techniques in material science specially nano-materials and explore approaches to maximize the potential of such techniques to be deployed in real-world applications. The approach proposed (SEM-Net) aims for the feature reduction with the Binary Particle Swarm Optimization method to execute the classification process on SEM images by concatenating the deeper layers of pre-trained CNN models AlexNet (fc6) and ResNet-50 (avg_pool). The best accuracy value was observed with 3112 features for the SEM dataset with optimized vectors as 99.3 %.The work aims to develop a new system can easily classify nanoparticles types using recent deep learning techniques. The main objective of the proposed system is giving the highest possible accuracy in nanoparticles features detection and classification.

Keywords

Artificial intelligence Material science Nanoparticles Nanoscience

Status: Abstract Accepted (Poster Presentation)