Winter School on AI: Course Syllabus & Schedule
Day 1: Wednesday, 6th January
Session 1: Unsupervised ML 1 – Foundations of Python & Machine Learning
- Time: 1:30 PM UK Time
- Duration: 1 hour 45 minutes
- Speaker: Soumya Banerjee
- Session Description:
This foundational module establishes the requisite Python environment and core programming concepts for machine learning before delivering an overview of unsupervised learning paradigms, algorithmic concepts, and target use cases.
- Learning Outcomes:
- Understand foundational Python syntax and data handling concepts.
- Differentiate between supervised and unsupervised learning approaches.
- Identify core unsupervised paradigms, including clustering and dimensionality reduction.
- Practical Activities: No (Theory and conceptual demonstration).
Session 2: Unsupervised ML 2 – Data Normalization, Dimensionality Reduction & Visualization
- Time: Following Session 1
- Duration: 1 hour 30 minutes
- Speaker: Soumya Banerjee
- Session Description:
An interactive workshop focused on data preprocessing, visual data exploration, and anomaly detection. Participants directly apply dimensionality reduction techniques to explore high-dimensional data structures.
- Learning Outcomes:
- Normalize raw datasets appropriately for machine learning models.
- Diagnose dataset errors, anomalies, and structural issues visually using Python.
- Implement Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) to visualize high-dimensional data.
- Practical Activities: Yes (Hands-on Workshop)
- Exercise 1: Preprocessing and normalizing dataset features in Python.
- Exercise 2: “Spot what is wrong” diagnostic exercise to identify data trends and anomalies.
- Exercise 3: Implementing PCA and t-SNE in Python to explore and visualize dataset structures.
Day 2: Thursday, 7th January
- Time: 1:30 PM UK Time
- Duration: ~2 hours
- Speaker: Course Faculty
- Session Description:
This session transitions from tabular unsupervised learning to modern natural language processing. It covers the core architectural principles behind Large Language Models (LLMs) and transformers, concluding with hands-on implementation.
- Learning Outcomes:
- Understand the architectural shift from traditional unsupervised learning to transformer-based LLMs.
- Grasp core transformer mechanics, including tokenization, embeddings, and self-attention.
- Load, configure, and run a basic transformer model using Python libraries.
- Practical Activities: Yes (Hands-on Workshop)
- Exercise 1: Practical execution of basic transformer models in Python.