1  Introduction

1.1 The ML Pipeline for Chemistry

The majority of ML projects in chemistry follow the same pipeline:

\[ \textit{Chemical system} \xrightarrow{\text{representation}} \mathbf{X} \xrightarrow{\text{model } f} \hat{y} \quad \text{(target: } y\text{)} \]

1. 💾 Data: collect, curate, and manage chemical data
2. 🧪 Representation: convert molecules, reactions, …, into machine readable representations \(\mathbf{X}\)
3. 🔧 Model: choose and train a model \(f\) that maps \(\mathbf{X}\) to predictions \(\hat{y}\) of a target property or label \(y\)
4. ✅ Validation: evaluate whether the model generalizes and whether \(\hat{y}\) matches the target \(y\) in a meaningful way
5. 🚀 Application: use models for prediction, interpretation, generation, or optimization

1.2 Course Overview

Lecture	Topic	Chapter
1	👋 Introduction	1
2	🧬 Cheminformatics, fingerprints, and descriptors	2
3	📈 Classical machine learning	3
4	🧐 Model selection and validation	4
5	🗃️ Data extraction, curation, and management	5
6	⚛️ Physics-based simulations and representations	6
7	🧠 Advanced machine learning	7–8
8–9	🕸️ Graph models	9
8–9	🔤 Language models	10
10	🤖 Explainable AI.	11
10	🔍 Unsupervised learning	12
11	🎨 Generative models for molecular design	13
12	🎯 Optimization problems	14

The relevant exercises from the course will be integrated here at a later point.