• Software Developers, Data Scientists, Entrepreneurs, Academics and Researchers

• Students should have some experience with Machine Learning. Familiarity with TensorFlow, Keras, and scikit-Learn is helpful but not mandatory.

Course Outline:

Representing Text for similarity            

• Word and Character Embeddings
• Keras Embedding Layers
• Word2Vec, CBOW and Skip-Gram Architecture
• Doc2Vec and Paragraph Vectors. PV-DM and PV-DBOW Architectures
• Ranking documents
• Capstone: Learning to rank

Classification in Text

• Neural Networks
• Backpropagation through time
• Capstone: Sentiment Analysis

Generating Text for suggestions

• Recurrent Neural Networks
• GRU and LSTM for memory retention
• Character and word level text generation
• Autocomplete and Autosuggestions
• Capstone: Writing like Shakespeare

Named Entity Recognition (NER) extracting knowledge

• SpaCy and NER Pipelines
• Bidirectional LSTM networks
• Capstone: Translating
  

Language Modeling with Attention                           

• Self-Attention and “Hey Siri”
• Transformers
• Using BERT and GPT-2
• Knowledge Distillation of language models
• Capstone: Language models in proteins
  

Hackathon

• Distilling BERT for detecting toxic comments. 

Software Required

This Practical NLP course is taught using Python >3.5 and  Tensorflow > 2.0 and Keras > 2.0 within Google Collaboratory. 

Training material provided: Yes (Digital format)