Welcome to the placeholder repository for AutoMatOOR, a Python package designed to democratize and centralize scalable solutions for automated materials optimization. This repository will be updated with the full package upon its official publication.
For inquiries, please contact Shayan Mousavi Masouleh at shayan.mousavimasouleh@nrc-cnrc.gc.ca.
Recent advances in machine learning and artificial intelligence are revolutionizing materials science, particularly in automated, self-driving laboratories. These labs perform experiments with minimal human oversight, pushing the boundaries of scientific research. AutoMatOOR addresses the challenge of developing customized optimization pipelines for unique research initiatives, which limits scalability. It provides a centralized, scalable platform to simplify complex, data-driven workflows and experimental designs.
-
Data Analysis Module: Offers a comprehensive statistical dashboard with conventional methods like correlation analysis and advanced techniques such as noise estimation and data quality assessment.
-
Predictive Modeling Module: Utilizes machine learning models to forecast experimental outcomes. It integrates AutoML frameworks to autonomously train and benchmark predictive models.
-
Active Learning and Design of Experiment Module: Optimizes research efficiency by refining experiment designs based on real-time feedback, significantly reducing the number of required experiments.
AutoMatOOR is adaptable, offering numerous analytical tools and pre-defined experimental pipelines. It supports multiple operational configurations including manual, human-in-the-loop, AutoML-driven, and intelligent agent-based modes.
Methodologies developed with domain experts and extracted from scientific publications are stored using Retrieval-Augmented Generation (RAG) and Cached Augmented Generation (CAG) algorithms. This aids users in selecting, customizing, and integrating pipelines effectively.
Stay tuned for the official release of AutoMatOOR, which will empower scientists across various expertise levels, fostering advancements in the materials science domain.