1 Machine Learning in Materials Processing & Characterization
4th/5th Semester – 5 ECTS · 2h lecture + 2h exercises per week
Coordinated with “Mathematical Foundations of AI & ML” (MFML)
and “Materials Genomics” (MG)
1.1 Synergy Map
- MFML provides the mathematical spine: loss functions, neural networks, generalization, uncertainty, Gaussian Processes.
- This course (ML-PC) applies these concepts to experimental data: images, spectra, and processing signals.
- Materials Genomics focuses on crystal structures, databases, and discovery.
ML-PC is therefore application-driven, not algorithm-driven.
1.2 Week-by-Week Curriculum (14 weeks)
1.2.1 Unit I — Experimental Data as a Learning Problem (Weeks 1–3)
1.2.1.1 Week 1 – What makes materials data special?
- Types of experimental data: micrographs, EBSD, EDS, EELS, XRD, process logs, thermal histories.
- PSPP (Processing–Structure–Property–Performance) as a data dependency graph.
- Why ML failure modes are common in experimental science.
Exercise:
Inspect real microscopy and process datasets; identify sources of bias and noise.
1.2.1.2 Week 2 – Physics of data formation
- Image and signal formation in characterization: resolution, contrast, artifacts.
- Sampling, aliasing, noise as physical priors (not preprocessing tricks).
- Relation to MFML refresher on PCA and covariance.
Exercise:
Fourier inspection of micrographs; effects of sampling and filtering.
1.2.1.3 Week 3 – Data quality, labels, and leakage
- Annotation uncertainty and inter-annotator variance.
- Train/test leakage in materials workflows.
- Why “good accuracy” often means a broken pipeline.
Exercise:
Construct a deliberately flawed ML pipeline and diagnose its failure.
1.2.2 Unit II — Representation Learning for Microstructures (Weeks 4–6)
(Aligned with early neural networks in MFML)
1.2.2.1 Week 4 – From classical microstructure metrics to learned representations
- Grain size, phase fractions, orientation maps.
- Limits of hand-crafted microstructure features.
- Transition to learned representations.
Exercise:
Compare classical features vs simple NN-based features for microstructure tasks.
1.2.2.2 Week 5 – Neural networks for microstructure images
- CNN intuition: filters as structure detectors.
- Example tasks: phase segmentation, defect detection, porosity identification.
- Overfitting risks with small datasets.
Exercise:
Train a small CNN on microstructure images; analyze failure cases.
1.2.2.3 Week 6 – Data scarcity & transfer learning
- Why materials datasets are small.
- Transfer learning from natural images vs self-supervised pretraining.
- When transfer learning helps—and when it does not.
Exercise:
Fine-tune a pretrained model; compare against training from scratch.
1.2.3 Unit III — Learning from Processing Data (Weeks 7–9)
1.2.3.1 Week 7 – Time-series and process monitoring
- Processing signals: temperature cycles, AM melt pool signals, SPS, rolling.
- Regression and sequence models as surrogates.
- Relation to MFML concepts of generalization.
Exercise:
Predict a process outcome from time-series data using regression or simple RNNs.
1.2.3.2 Week 8 – Generalization, robustness, and process windows
- Sensitivity to noise and parameter drift.
- Overfitting in process–property models.
- Robustness as a design criterion.
Exercise:
Analyze model robustness under perturbed process conditions.
1.2.3.3 Week 9 – Inverse problems and process maps
- Process → structure inverse problems.
- ML-guided process maps (e.g. AM laser power vs scan speed).
- Physics-informed vs unconstrained regression.
Exercise:
Construct a simple ML-based process map; compare constrained vs unconstrained models.
1.2.4 Unit IV — Uncertainty, Surrogates, and Automation (Weeks 10–12)
1.2.4.1 Week 10 – ML for characterization signals
- Spectral data: XRD, EELS, EDS.
- Denoising, peak finding, dimensionality reduction.
- Using ML without destroying physical meaning.
Exercise:
Apply PCA/NMF to spectral datasets; interpret components physically.
1.2.4.2 Week 11 – Automation in microscopy and characterization
- Autofocus, drift correction, parameter selection.
- ML as a control component, not just a predictor.
Exercise:
Implement a simple ML-assisted autofocus or defect detector.
1.2.4.3 Week 12 – Uncertainty-aware regression & Gaussian Processes
- Aleatoric vs epistemic uncertainty in experiments.
- Gaussian Processes as uncertainty-aware surrogates.
- Exploration vs exploitation in experimental design.
- Connection to materials acceleration platforms.
Exercise:
Compare GP regression and NN ensembles for a process-parameter problem.
1.2.5 Unit V — Physics, Trust, and Synthesis (Weeks 13–14)
1.2.5.1 Week 13 – Physics-informed and constrained ML
- Embedding physical constraints into ML models.
- Penalty terms, soft constraints, hybrid approaches.
- Failure modes of unconstrained models.
Exercise:
Train a constrained model for a processing or characterization task.
1.2.5.2 Week 14 – Integration, limits, and reflection
- Explainability for experimental ML (CAMs, SHAP).
- Why ML fails in real labs.
- Where ML genuinely changes materials processing.
Exercise:
Mini-project presentations and critical discussion.
1.3 Learning Outcomes
Students completing this course will be able to:
- Interpret materials processing and characterization data as learning problems.
- Build ML pipelines for microstructure analysis, process prediction, and spectral data.
- Understand the physics of data formation to avoid common ML pitfalls.
- Evaluate generalization, robustness, and uncertainty in experimental ML models.
- Apply Gaussian Processes and neural networks as surrogate models.
- Integrate physical constraints into ML workflows.
- Critically assess claims about ML in materials processing and characterization.
1.4 Lab Possibilities
- Microscopy datasets: noise, metadata, units, and artifacts.
- Fourier inspection of SEM/TEM images.
- Broken vs correct ML pipelines (data leakage case studies).
- Feature extraction vs learned representations.
- Fine-tuning pretrained CNNs on microstructures.
- Process–property regression with uncertainty.
- GP-based process maps.
- Spectral decomposition (NMF) of EELS/XRD data.
- ML-assisted autofocus or EBSD pattern classification.
- Multi-modal fusion of images, spectra, and process parameters.