WORKSHOP
“Machine learning for scientists: supervised, unsupervised , and probabilistic learning”
We bring together leading academic scientists and researchers!
September 4th, 2026 1pm - 5pm
Scope of the event
Machine learning has entered the vocabulary of every branch of physical science, yet its relationship to es- tablished practices of statistical modelling and scientific inference is often poorly articulated. This opening workshop addresses that gap directly. It establishes what machine learning is and is not, situating the field within the broader landscape of data analysis that physical scientists already practise, and drawing a careful distinction between fitting a model to data and learning generalisable structure from it.
The workshop introduces core concepts — loss functions, regularisation, overfitting, cross-validation, and the bias–variance trade-off — through examples drawn from spectroscopy, thermodynamics, and materials prop- erty prediction rather than from image classification or commercial recommender systems. Supervised learning (regression and classification), unsupervised learning (clustering, dimensionality reduction, latent-variable mod- els), and probabilistic framing (likelihood, Bayes’ theorem, posterior predictive distributions) are developed with attention to the kinds of data, noise models, and physical constraints that characterise experimental and computational science.
The session also addresses model selection and validation methodology — train/validation/test splits, k-fold cross-validation, and learning curves — emphasising the pitfalls that arise when small, expensive datasets are involved, as is typical in physical science. Participants will gain a clear conceptual foundation for the methods developed in subsequent workshops, together with practical guidance on when ML adds genuine value to a scientific investigation and when simpler or more interpretable approaches are preferable.
Extended Key Takeaways & Outputs
• Clear conceptual framework for supervised, unsupervised, and probabilistic ML in a scientific context
• Practical understanding of model selection and validation for small, expensive datasets
• Identification of common failure modes: overfitting, data leakage, and inappropriate generalisation claims
• Guidance on when ML adds value and when conventional analysis suffices
• Foundation for all subsequent workshops in the series
Agenda & Speakers
13:00–13:05 — Introduction & Objectives
Organizer
· Positioning machine learning within scientific data analysis and modelling
· Distinction between statistical inference and machine learning
· Workshop objectives: clarity, rigour, and applicability to physical sciences
13:05–13:35 — Supervised Learning: Regression and Classification for Scientific Data
Speaker (Academia): N.N.
· Regression models for predicting physical and chemical properties
· Classification methods for experimental and simulation datasets
13:35–14:05 — Unsupervised Learning: Structure Discovery in Scientific Data
Speaker (Academia): N.N.
· Clustering and pattern recognition in high-dimensional datasets
· Dimensionality reduction and latent-variable models
14:05–14:35 — Probabilistic Machine Learning and Bayesian Inference
Speaker (Academia): N.N.
· Likelihood-based modelling and uncertainty quantification
· Bayesian inference and posterior predictive distributions
14:35–14:45 — Break
14:45–15:15 — Machine Learning in Industrial R&D: From Data to Insight
Speaker (Industry): N.N.
· Integration of ML into experimental workflows and materials discovery
· Challenges of data quality, scale, and reproducibility
15:15–15:45 — Model Validation and Generalisation in Scientific Applications
Speaker (Industry): N.N.
· Cross-validation, learning curves, and performance evaluation
· Avoiding overfitting and data leakage in small datasets
15:45–16:15 — When to Use (and Not Use) Machine Learning in Science
Speaker (Policy/Applied Research): N.N.
· Trade-offs between model complexity and interpretability
· Role of physics-based models and simpler analytical approaches
16:15–16:55 — Panel Discussion & Strategic Alignment
All Speakers + Moderator
· Integrating machine learning into the scientific method
· Ensuring robustness, transparency, and reproducibility
· Building collaboration between domain scientists and ML experts
Concluding remarks
Registration
Deadlines:
Early-Bird: until July 15th, 2026
Registration : until September 5th, 2026
The registration takes place online by filling out the form.
The registration and access to the virtual room for workshop is effective only if the fees payment has been confirmed.
Please note that INVIRTA MEMBERS can purchase all our workshops and services with 10% discount . If you desire to get more information on our membership please press here .
Prices* :
Industry: 150 € (early bird**: 125 €)
Academy: 125 € (early bird**: 100 €)
Students***: 90 € (early bird**: 75 €)
* Note that prices are subject to VAT (for European countries, except Germany, provide the VAT number of your organisation to be within the VAT reverse framework payment).
** Until November 15th, 2020
*** Copy of the registration certificate (Ph.D, undergraduates)
Special price offers for Workshop packages
Our Workshop packages are valid for any workshops offered by our academy and can be used anytime. If you want to use one of our packages press here.
