4-Day Machine Learning Course for Life Sciences

Materials of the course delivered by Machine Learning for Biomedical Research Unit at Barcelona Supercomoputing Center at May 2025.

• Day 1: Data Processing & Feature Engineering — Python refresher, data exploration, preprocessing, and feature engineering.

• Day 2: Supervised & Unsupervised Learning — Models like linear/logistic regression, tree-based methods, clustering, plus hands-on training.

• Day 3: Deep Learning — From theory to practice with PyTorch & fast.ai; image, text, and tabular applications.

• Day 4: Large Language Models — Transformer theory, Hugging Face demo, and fine-tuning on life sciences NLP tasks.

💡 Hands-on sessions via Google Colab. No local setup required.

Day 1 – Data Processing and Feature Engineering

• Python Refresher (pandas, numpy, plotting, scikit-learn)
• Data Exploration
• Data Preprocessing
• Feature Engineering (Extraction & Selection)
• Hands-on session

Notebook Title	Open in Colab
Introduction to Python
Features Engineering

Day 2 – Supervised and Unsupervised learning

• Linear Regression, Logistic Regression, Tree-Based Methods
• Random Forest, XGBoost Unsupervised Learning (1 h)
• K-means, DBSCAN
• Hands-on: Training and Testing Supervised and Unsupervised Models

Notebook Title	Open in Colab
Supervised Learning
Unsupervised Learning

Day 3 – Deep Learning

• Understand key differences between Machine Learning and Deep Learning
• Learn the backpropagation algorithm and how neural networks learn
• Dive into Convolutional Neural Networks (CNNs) and their applications
• Explore PyTorch and fast.ai for building and training deep models
• Hands-on projects: image classification, segmentation, text and tabular data, and recommendation systems
• SHAP values for model explainability and interpretation

Notebook Title	Open in Colab
Gradient Descent demo
fast.ai intro
SHAP
PyTorch Intro

Day 4 – Large Language Models from Theory to Applications

• Introduction to core concepts behind Large Language Models (LLMs)
• Explanation of Transformer architecture and attention mechanisms
• Discussion of strengths, limitations, and real-world scaling challenges
• Hands-on demo with Hugging Face Transformers library
• Practical exercise: fine-tuning an LLM on a downstream task

Notebook Title	Open in Colab
Hugging Face Demo
Translation Lab

Further steps - learning materials

🎓 Free Courses to Start With

Course Name	Description	Link
Deeplearning.ai Machine Learning Specialization	For sequential learners, who prefer to start from basic ideas and progress to complex ideas	Link
fast.ai - Practical Deep Learning	For top-down learners, who prefer to start from examples and working code	Link

---

🛠️ Advanced Concepts, Engineering

Course Name	Description	Link
Full Stack Deep Learning - MLOps	How to put Deep Learning models into production	Link
DataTalksClub	Set of courses on Data Engineering, MLOps, LLMs. Focus on engineers	Link

---

🤖 LLMs

Course Name	Description	Link
Hugging Face LLM Course	Comprehensive LLM course by Hugging Face	Link
Hugging Face Smol Course	Hugging Face's beginner-friendly course	Link
Hugging Face Agents Course	Learn how to use agents with Hugging Face tools	Link
Hugging Face MCP Course	Multi-modal Course Pack focused on multiple modalities	Link