- Create and train deep neural network architectures (such as Convolutional Neural Networks, Recurrent Neural Networks, LSTMs, Transformers) and apply them to tackle NLP applications such as text classification, machine translations, information extraction, text simplification and text summarization. - Study, implement and use word embeddings to represents texts.

1. Introduction 2. NLP basic tasks. 3. Word embeddings. 4. Deep Learning architectures for text classification. 5. Deep Learning architectures for named entity recognition (NER). 6. Deep Learning architectures for relation extraction (RE). 7. Text similarity 8. Deep Learning architectures for text summarization. 9. Deep Learning architectures for machine translation.

The flippled classroom model will be applied: Each week, the teacher will publish in aulaglobal a Jupyter notebook that describes and contains the implementation of a deep architecture for the resolution of a certain NLP application (for example, the detection of fake news or the anonymization of clinical notes). Students should study and run these notebooks before class. During class, the teacher will explain the notebook and resolve any possible doubts. Once resolved, the teacher will propose possible improvements and raises new challenges or problems to solve. Students should try to extend or adapt the implementations of notebooks to solve the new challenges posed by the teacher. The methodology is practical. The classes will have a configuration of work in pairs in a computer room. Tutoring by appointment is limited to 2 hours per week. These can be online or face to face.