Machine Learning for Experimentation does not have to be hard
Matterhorn streamlines the integration of Machine Learning in your experimentation process
Matterhorn in 3 steps
This 5 minute tutorial demonstrates the three fundamental steps of experimental material discovery:
- Upload data and specify variables
- Build Machine Learning models for estimation
- Calculate the next best experiment with Bayesian Optimisation
Get a feeling for what data-driven material development can feel like. Understand the advantages and limitations of Machine Learning and learn how Bayesian Optimisation can work in your laboratory.
Start TutorialBlog
(Podcast) Welt der Werkstoffe - talk, Folge 24, Jakob Zeitler: Einsatz von KI in der Werkstoffforschung
I had the chance to talk about Machine Learning in Material Science with Professor Bonnet on his show "Welt der Werkstoffe".
Feb. 1, 2023
(Paper Review) A Self-driving Laboratory Optimizes A Scalable Process For Making Functional Coatings
I had the chance to discuss with Connor Rupnow his paper on a SDL that optimises functional coatings and learned a lot about the fundamental issues of implementation.
Jan. 2, 2023
(Workshop) AI4Mat NeurIPS 2022 Workshop
On December 2, 2022, NeurIPS 2023 hosted the 1st workshop on AI for Accelerated Design (AI4Mat). The goal was to bring together researchers and domain experts from both AI and materials science to collaborate on open problems.
Dec. 5, 2022
-
01 How much data do I need?
The more the better! Most labs will already have a dataset to build from. Importantly, the data needs to be of high quality, i.e. low noise. Ideally, it is composed of continuous variables and few dimensions, as categorical variables are generally harder to optimise.
-
02 Do I need to know any Machine Learning?
Matterhorn is indifferent to how much Machine Learning capabilities your lab currently offers. Feel free to bring your own Data Scientist, but if you are in the early stages of ML-driven material science we are happy to support you with recruiting or providing consultation.
-
03What is Bayesian Optimisation for Material Design?
Bayesian Optimisation is an algorithm that guarantees the statistically most efficient path to optimal material performance. It uses the uncertainty in the solution space to decide which experiment to run next.
Measurable impact on your experimentation workflow
See paper Autonomous experimentation systems for materials development
Time up to 10x faster
Cost up to 100x cheaper
Success up to 5x higher success rate
Features
Material discovery requires several stepping stones. Matterhorn takes care of your data management and visualisation needs. User-friendly modeling and optimisation interfaces seamlessly integrate in any laboratory workflow.
Data Management
Manage your data and variables in a single storage for efficient exchange across your team
Optimisation
Schedule your next experiment, efficiently. Choose from a variety of search strategies for optimal success rate.
Get in touch
We will happily guide you through the emerging space of data-driven material discovery. We look forward to learn from your experience and problem space.
