Here is a brief diary of topics discussed in classes.
Slides and coding scripts are distributed after each lecture via the Google groups mailing-list. Refer to the home page on how to subscribe.
[25.09.2024] Introduction to the course, review introduction of machine learning and computer computer vision.
[27.09.2024] Review of the basics of digital image processing and of neural networks, specifically fully connected layers, the computational graph, backpropagation.
[2.10.2024] Review of Pytorch basics including multi-layer perceptons.
[4.10.2024] Pytorch practical including ConvNets and hyper-parameter tuning.
[9.10.2024] Convnets.
[11.10.2024] Visualization of ConvNets, Activation functions, data pre-processing.
[16.10.2024] Introduction of the assignment, graph meta networks and machine unlearning.
[18.10.2024] Weight initialization, batch normalization, DropOut and data augmentation.
[23.10.2024] Hyper-parameter tuning; transfer learning, introduction of sequence modelling
[25.10.2024] Sequence modelling with temporal convolutional networks and recurrent neural networks.
[30.10.2024] Backprop in RNNs, LSTMs and image captioning with RNNs.
[06.11.2024] Image captioning with Attention
[08.11.2024] Self-attention, Positional encoding and Transformer Networks.
[13.11.2024] Transformer Networks for images and BERT.
[15.11.2024] BERT and the deployment of large-scale models with Knowledge Distillation; Introduction of final projects.
[20.11.2024] Applications of sequence modelling to earthquake forecasting and to human trajectory forecasting.
[22.11.2024] Graph encoders and Graph Neural Networks.
[27.11.2024] Graph Convolutional Networks and applications to human pose forecasting.
[29.11.2024] First project presentations.
[4.12.2024] Semantic segmentation.
[6.12.2024] Object Detection and Instance Segmentation.
[11.12.2024] Multi-modal Learning.
[18.12.2024] Generative AI models, including auto-encoders, variational auto-encoders, diffusion models, and application to text-to-model generation.