Subject Overview

This subject develops foundational and professional competence in deep learning and natural language processing (NLP). It combines neural-network theory, language representation, contemporary architectures, model development, deployment, analytics, governance and responsible business application.

Learning Goal

Students will be able to explain core deep learning and NLP concepts; prepare language data; build, evaluate and deploy neural models; use pretrained and generative AI responsibly; interpret model outputs; and support evidence-based organizational decisions.

Syllabus

1. Foundations of Deep Learning and NLP

Learning Objective: Explain artificial intelligence, machine learning, deep learning and NLP; distinguish major learning paradigms; identify language-processing tasks, terminology, industry use cases and limitations.

Category: foundation

2. Mathematical and Computational Foundations

Learning Objective: Apply essential vectors, matrices, probability, statistics, derivatives, loss functions and gradient-based optimization; use Python, notebooks and tensor frameworks for reproducible experiments.

Category: foundation

3. Neural Networks and Learning Processes

Learning Objective: Describe perceptrons, multilayer networks, activation functions, forward propagation, backpropagation, initialization, regularization, normalization, hyperparameters and overfitting control.

Category: core

4. Language Data Acquisition and Preparation

Learning Objective: Collect, label and manage text corpora; perform cleaning, normalization, tokenization, stemming, lemmatization and subword segmentation while addressing data quality, imbalance and leakage.

Category: core

5. Text Representation and Language Features

Learning Objective: Compare one-hot encoding, bag-of-words, n-grams, TF-IDF, static embeddings and contextual embeddings; select suitable representations using semantic and computational criteria.

Category: core

6. Classical NLP and Baseline Models

Learning Objective: Apply linguistic concepts and conventional methods to part-of-speech tagging, parsing, named-entity recognition, topic analysis, text classification and information retrieval; establish meaningful baselines.

Category: core

7. Sequence Models for Language

Learning Objective: Explain and implement recurrent neural networks, LSTM, GRU, bidirectional models, encoder-decoder architectures, attention and sequence-to-sequence learning for language tasks.

Category: core

8. Transformers and Pretrained Language Models

Learning Objective: Explain self-attention, positional encoding, transformer components and model families such as BERT and GPT; apply transfer learning, fine-tuning, parameter-efficient adaptation and prompt design.

Category: core

9. Applied NLP Systems

Learning Objective: Develop solutions for sentiment analysis, document classification, search, question answering, summarization, translation, conversational agents and information extraction using appropriate pipelines and models.

Category: application

10. Large Language Models and Generative AI

Learning Objective: Use large language models through APIs or open-source tools; apply prompting, embeddings, vector databases and retrieval-augmented generation; recognize hallucination, context and grounding limitations.

Category: contemporary

11. Model Evaluation, Interpretation and Error Analysis

Learning Objective: Use suitable train-validation-test designs and metrics, including precision, recall, F1, ROC-AUC, perplexity, BLEU and ROUGE; conduct qualitative error analysis, explainability checks and robustness testing.

Category: application

12. Deployment, MLOps and Production Monitoring

Learning Objective: Package and serve NLP models; manage versions, experiments and pipelines; assess latency, scalability, cost and privacy; monitor drift, quality and failures throughout the model lifecycle.

Category: application

13. Responsible AI, Ethics and Governance

Learning Objective: Evaluate bias, fairness, privacy, security, intellectual property, transparency and environmental impact; apply human oversight, documentation, risk controls and relevant AI governance principles.

Category: strategic

14. Business Integration and Emerging Directions

Learning Objective: Translate organizational problems into NLP projects; assess feasibility, value, risk and build-versus-buy choices; support analytics and decision-making while examining multimodal AI, efficient models, agents and future workforce implications.

Category: contemporary