Labs from the course ALTEGRAD (Advanced Learning for Text and Graph Data)
Check this link !
Lab 1 : Neural Machine Translation
- Sequence to Sequence (seq2seq) architectures.
- Implementation of the Neural Machine Translation (NMT) model from This paper
- Study of Recurrent Neural Networks
- Study of Global Attention Mechanism
Lab 2 : Graph Mining
- Dynamics of a Real-World Graph.
- Community Detection / Clustering; spectral clustering, Modularity.
- Graph Classification usign Graph Kernels.
Lab 3 : Transfer Learning in NLP
- Generative pre-training of a language model.
- Transformer Model.
- Vocabulary and Tokenization.
Lab 4 : NLP Frameworks
- Pretrain and finetuning of transformer based language models
- Fairseq / LoRa (Low-Rank Adaptation)
- ROBERTaSmall Model
Lab 5 : Deep Learning for Graphs 1/2
- Node Embeddings; DeepWalk.
- Graph Neural Networks; Implementation, Node Classification.
Lab 6 : Deep Learning for Graphs 2/2
- Graph Neural Networks; expressive power of GNNs.
Lab 7 : Learning on Sets and Graph Generative Models
- DeepSets
- Graph Generation with Variational Graph Autoencoders