Week 8 - Natural Language Processing & Deep Learning
Week: 2025-10-27 to 2025-10-31
Status: “Done”
Week 8 Overview
This week provides a comprehensive deep-dive into Natural Language Processing (NLP), covering linguistic foundations, modern NLP applications, sequence-to-sequence architectures, and evaluation methodologies. From understanding phonetics to implementing machine translation systems, this week bridges theory and practice in NLP.
Key Topics
Linguistic Foundations
- Core components: Phonetics, Phonology, Morphology, Syntax, Semantics, Pragmatics
- Understanding how language structure informs NLP design
NLP Applications
- Search engines and intent recognition
- Online advertising with NER and relationship extraction
- Voice assistants and speech recognition
- Chatbots with NLU/NLG pipelines
- Machine translation systems
- Text summarization (extractive & abstractive)
Deep Learning Architectures
- Seq2seq models with encoder-decoder architecture
- LSTM deep dive: forget gate, input gate, cell state, output gate
- Attention mechanism and self-attention
- Neural Machine Translation (NMT) implementation
Evaluation & Decoding
- BLEU score (precision-based)
- ROUGE score (recall-based)
- F1 score for MT evaluation
- Beam search decoding
- Minimum Bayes Risk (MBR) sampling
Hands-on Labs
- Building voicebot and chatbot workflows
- Implementing LSTM for sequence modeling
- Creating encoder-decoder with attention
- Neural machine translation end-to-end
- Evaluating translation quality with BLEU/ROUGE
- Implementing beam search and MBR