Chapter 14: Deep Learning II — Sequences, NLP & RL Vocabulary
Deep Learning II — Sequences, NLP & RL Vocabulary in ML Software Engineering: Interview Concept Review.
Learning Objectives
By the end of this chapter, you will be able to:
- Relate Deep Learning II — Sequences, NLP & RL Vocabulary to common ML software engineering interview questions and trade-offs.
- Explain when this topic deserves a deeper pass through another tutorial on this site versus staying at recap depth.
- Surface assumptions, pitfalls, and follow-up probes an interviewer is likely to use.
Discrete tokens → vectors
Word2Vec / GloVe: local window co-occurrence vs global statistics; both yield static embeddings lacking polysemy richness—contextual models supersede for production NLU—but interviews still probe reasoning.
RNNs recurse hidden state across time; exploding/vanishing gradients motivate LSTM/GRU gating (forget/input/output structures verbally).
Self-attention mixes all positions with softmax weights—parallelizable unlike recurrent cores; dominates modern LLMs. Tie to quadratic memory in sequence length caveat.
MDP / Q-learning snapshot
MDP tuple (states, actions, transitions, rewards, discount). Q-learning learns action values; policy gradient alternatives exist—keep story honest if role not RL-heavy.
Go deeper on this site
1. Transformer self-attention parallelism advantage vs recurrence?