21AI62 Dayananda Sagar University • CSE — Artificial Intelligence & Machine Learning (Semester 6)

Natural Language Processing

💡 DSU Exam Strategy & Guidance

Text Preprocessing (Tokenization, Lemmatization), N-gram Language Models, Word Embeddings (Word2Vec, GloVe), Sequence Tagging (HMM/CRF), and Pretrained LLMs (BERT, GPT).

High-Yield Passing Strategy: Explain Skip-Gram vs Continuous Bag of Words (CBOW) in Word2Vec and BERT Masked Language Modeling.
Most Repeated Question Topics:
Word2Vec Skip-Gram Architecture with Negative SamplingBERT vs GPT Architectural DifferencesN-gram Language Modeling & Smoothing TechniquesTF-IDF & Cosine Similarity for Semantic Search

🔥 Most Predictable Exam Questions (1)

95% Probability 8 Marks Module 4: Sequence Models & Transformers

Q1. Explain the Transformer Self-Attention mechanism mathematically: Attention(Q, K, V) = softmax((Q K^T) / sqrt(d_k)) * V and why scaling factor sqrt(d_k) is required.

Key Points: Define Queries (Q), Keys (K), Values (V) as linear projections of input embeddings. Dot product Q K^T measures semantic similarity. sqrt(d_k) scaling prevents dot product values from growing extremely large for large dimensions, which would push softmax into regions with vanishing gradients.

🎓 Open Source Courses & Video Playlists (2)

Stanford Online

Stanford CS224N: NLP with Deep Learning (Prof. Christopher Manning)

Watch Course ↗
Hugging Face

Hugging Face NLP Course

Watch Course ↗

📄 Previous Year Question Papers (1)

END_SEM • 2024

End Semester Exam 2024

View PDF ↗

Turn 21AI62 Notes into AI Flashcards

Upload your Natural Language Processing lecture slides to generate spaced repetition study cards in seconds.

Start Free →
MindFlow
DSU Hub Sign In

Structured Active Recall for Complex Coursework.

Turn your lecture PDFs into flashcards, adaptive quizzes, and spaced repetition reviews in seconds.

Start Free Sprint →