Session 1: Introduction + Linear models (regression)

• Nicole Hartman (TUM)

- Welcome + logistics - Intro AI, ML, and deep learning - Machine Learning: Key ideas - Linear models (regression) - Bias variance trade off **Bonus:** Linear algebra review: https://cs229.stanford.edu/notes2022fall/cs229-linear_algebra_review.pdf

Session 2: Perceptrons and Multi-layer perceptrons

• Nicole Hartman (TUM)

- Perceptron (Linear Model + Classification) - Multi-layer perception (MLP, a.k.a, Neural Network) - Universal Approximation Theorem - Forward pass: MLP (in numpy)

Session 3: How NNs learn

• Nicole Hartman (TUM)

- Automatic Differentiation (AD) - Computational Graphs and backpropogation - Code full NN from scratch (numpy)

Session 4: Deep Learning Frameworks and Optimizers

• Nicole Hartman (TUM)

- Intro to pytorch (autodiff framework) - Loss functions - Optimizers

Session 5: Going deeper... with tips and tricks

• Nicole Hartman (TUM)

- Increasing depth - Regularization tricks - Ensembling methods - Tips and tricks for training NNs

Session 6: Clustering + Unsupervised Learning

• Nicole Hartman (TUM)

- Softmax - PCA - K-means

Lecture day 4

• Baran Hashemi (ORIGINS Cluster Munich)

- Introduction to Geometric Deep Learning (GDL) - GDL on grids and Convolutional Neural Netowrks

Tutorial Day 4

• Baran Hashemi (ORIGINS Cluster Munich)

Lecture day 4

• Baran Hashemi (ORIGINS Cluster Munich)

- Geometric Deep Learning on Sets and Graphs: - Deep Sets and Graph Neural Networks I - Graph Neural Networks II

Tutorial Day 4

• Baran Hashemi (ORIGINS Cluster Munich)

Lecture day 4

• Baran Hashemi (ORIGINS Cluster Munich)

- Introduction to Deep Generative Models I

Tutorial Day 4

• Baran Hashemi (ORIGINS Cluster Munich)

- Introduction to Deep Generative Models II