Intro to Materials AI

ePotentia

Gain hands-on experience in Materials AI by completing two practical case studies on the Flemish Supercomputer, presented by ePotentia.

Sterre, S9,  Multimediazaal

Artificial Intelligence (AI) is becoming omnipresent in materials research as well as industrial production. The reason is its capacity to make optimal use of available data, which is only becoming more plentiful in times of industry 4.0.The hands-on session introduces you to Materials AI through two practical case studies. In each case study, special attention is given to explainable AI (xAI). xAI aims to return materials insight to models which can otherwise be difficult to interpret.The case studies focus on the topic of detecting defects on a material surface, in this case steel plates. In the first case study, you will learn how to solve this problem using tabular data, as typically found in a spreadsheet such as Excel. You will learn how to create classical machine learning models (random forests), optimize them on the HPC and interpret them using explainable AI. In the second case study the same problem will be treated with deep learning, more specifically convolutional neural networks. You will learn how to optimally train models on HPC GPU hardware, use data augmentation techniques to improve model quality and robustness, as well as regain insight using explainable AI.While we focus on a single application, detecting steel defects, the same AI techniques can be applied to any type of tabular or image data.This hands-on session is part of the AI4MI course project (<a href="https://ai4mi.epotentia.com" target="_blank" data-msys-clicktrack="0" rel="nofollow noopener noreferrer">https://ai4mi.epotentia.com</a>).The session will take place in an interactive environment on the Flemish Supercomputer, using the Python programming language. Example code will be provided in graphical notebook format, but it is recommended to learn basic Python before attending.<h3>Agenda</h3>10:00 – 13:00: Case study 1: Introduction to machine learning for tabular data (steel defects)13:00 – 14:00: Sandwich lunch (free)14:00 – 17:00: Case study 2: Introduction to deep learning for image data (steel defects)<h3>Organizers</h3>ePotentia is an AI consultancy company focused on scientific and industrial AI solutions. Active sectors range from materials industry to life sciences with experience in data management and analysis, model building, cloud deployment and app development.This session is funded by Research Foundation Flanders and the Flemish Supercomputer Centre. The original course was funded by the European Social Fund and OCAS NV.

Our {communityGuidelinesLink} describe the sort of content we prohibit on Eventbrite. If you suspect an event may be in violation, you can report it to us so we can investigate.

To report other categories of prohibited or illegal content, submit {link}