These pages contain learning resources and laboratory exercises for the module: Introduction to Deep Learning for Speech and Language Processing, delivered by Division of Psychology and Language Sciences, University College London.

All materials (c) 2020 Mark Huckvale

Tutorials

• Tutorial Python
• Tutorial Numpy
• Tutorial Matplotlib
• Tutorial Pandas
• Tutorial NLTK

Demonstrations

• RNN Demo - Demonstration of Recurrent Neural Networks
• LM Demo - Demonstration of Language Modelling
• Seq2Seq Demo - Demonstration of Sequence-to-sequence processing

Exercises

• Exercise 1.1 - Python language exercises [Answers]
• Exercise 1.2 - Numpy exercises [Answers]
• Exercise 1.3 - Matplotlib exercises [Answers]
• Exercise 2.1 - Pandas exercises [Answers]
• Exercise 2.2 - Regression problem [Answers]
• Exercise 2.3 - Classification problem [Answers]
• Exercise 3.1 - Gradient descent [Answers]
• Exercise 3.2 - Autoencoder [Answers]
• Exercise 3.3 - Vowel classifier [Answers]
• Exercise 4.1 - Sentiment analysis [Answers]
• Exercise 4.2 - Emotion classification [Answers]
• Exercise 4.3 - Prediction of speaker age [Answers]
• Exercise 5.1 - Word similarity in Wordnet [Answers]
• Exercise 5.2 - Training word embeddings [Answers]
• Exercise 5.3 - Using word embeddings for tagging [Answers]
• Exercise 6.1 - RNN for sentiment analysis [Answers]
• Exercise 6.2 - Isolated word speech recogniser [Answers]
• Exercise 6.3 - Calculation of phone posteriors [Answers]
• Exercise 7.1 - Character-level language model [Answers]
• Exercise 7.2 - Cloze task predictor using word language model [Answers]
• Exercise 8.1 - Grapheme to phoneme conversion [Answers]
• Exercise 8.2 - Sentence embedding encoder [Answers]
• Exercise 8.3 - Neural machine translation [Answers]
• Exercise 9.1 - Movie dialogue chatbot [Answers]

To open one of these notebooks in Colaboratory or Binder,