M.Sc. AI – Consolidated Study Notes (Semesters I–III)

Semester I – Core Courses #### CAI-CC-511 Mathematical Foundations of AI (credits $3$) - Preamble - Bridges linear algebra + statistics with AI algorithms. - Emphasises matrix manipulation, regression, regularisation, SVD, optimisation. - Prerequisites Linear algebra, discrete maths, numerical analysis. - Course Outcomes - Solve linear‐algebra problems, perform matrix arithmetic, factorisations, PCA, relate probability & regression to AI, explore propositional/predicate logic, grasp information theory. - Modules - I – Linear Algebra Basics - Vectors, norms, cosine similarity, orthogonality, subspaces, basis, rank, linear (in)dependence. - II – Matrices & Factorisations - Determinant, Hadamard product, linear transformations, types (identity, invertible, covariance). - Eigen-pairs, diagonalisation, SVD, PCA for dimensionality reduction. - III – Probability - Probability space, sum/product/conditional rules, Bayes; Binomial, Poisson, Normal, Uniform, Exponential, Gaussian dists. - IV – Linear Regression - Formulation, parameter estimation, MLE, over-fitting, MAP $=\text{MLE}+\text{regularisation}$. - V – Logic - Syntax/semantics of propositional & first-order logic, WFF properties, resolution, clause form, unification. - VI – Markov & Info Theory - Markov process/chain; entropy, cross-entropy, mutual information; applications in ML & KR.

#### CAI-CC-512 Algorithms – Complexity & Optimisation (credits $3$) - Preamble Iterative + recursive algorithm analysis, big-O, efficient numeric/geometric/combinatorial algorithms, complexity theory. - Prerequisites Data structures, Linear algebra. - Outcomes Analyse performance, apply recurrences, DP, B&B, greedy (MST), heuristic search, P vs NP, optimisation & network flows. - Modules - I Asymptotics, recurrence solving (iteration, tree, substitution, Master). - II Greedy: Kruskal, Dijkstra. Graph search: DFS, BFS, ID, Best/Beam, B&B, $A^*$
. - III Heuristics: generate-and-test, hill-climb, simulated annealing, AO*, CSP, means–ends. - IV Complexity: P, NP, NP-Hard/Complete; proofs (Clique, Vertex-Cover); approximation (bin-packing, graph-colouring). - V Optimisation taxonomy; classical vs advanced; global vs local optima. - VI Applications: missionaries-&-cannibals, string matching, TSP, robotic motion planning, crypto-arithmetic, network flow.

#### CAI-CC-513 Principles & Ethics of AI (credits $3$) - AI timelines, branches, harms, intellectual roots. - Ethics: need, codes, epistemic/tech/moral strategies; normative theories; ethics

$
e$ banning. - Domains: Self, Friend, Stranger, World; autonomy, responsibility, bias elimination, machine ethics, moral agency. - Professional aspects: AI orgs, codes for machines, transparency (FAST: Fairness, Accountability, Sustainability, Safety, Transparency). - Case study WHO principles for AI in health (autonomy, well-being, transparency, accountability, inclusiveness, sustainability).

#### CAI-CC-514 Principles of Computing (credits $3$) - Formal languages, automata, computability/decidability. - Modules: 1. Regular languages, closure, pumping lemma. 2. DFA, NFA (ε), equivalence, minimisation (Myhill–Nerode). 3. CFG, ambiguity, CNF, PDA (deterministic), pumping lemma for CFG. 4. Turing machines (multi-tape, NDTM, enumerators). 5. Computability: decidability, halting, UTM, reducibility, recursion theorem, PCP. 6. Applications: circuits, networking, regex in search, CFG in NLP, TM studies.

#### CAI-CC-515 Knowledge Representation & Reasoning (credits $3$) - Agent structures, production systems, state spaces. - KR structures: FOL, frames, semantic nets, scripts, conceptual dependency. - Reasoning: non-monotonic, fuzzy, rule-, case-, model-based; Bayes rule/networks. - Game playing: minimax, alpha-beta, iterative deepening. - Expert systems: components, development, MYCINE case.

#### CAI-CC-516 Theory of Computing Lab (credits $3$) - Warm-up (functions, DS, I/O), Numerical (matrix ops, probability, Markov), Algorithm cycles (greedy, graphs, complexity), Language & Automata (regex, parse trees, NFA→DFA). - Assessment $=30+50+20$ marks, record required. ### Semester I – Skill / Generic / Elective - CAI-SE-4B1 Entrepreneurship & Professional Development (credits $2$)

Communication, project planning, soft skills, SWOC, career skills (CV, GD, interview), entrepreneurial traits & life cycle. - CAI-SE-4B2 IT Act & Constitution of India (credits $2$) (covers cyberspace law, constitutional rights, governance). - CAI-GC-4B1 Artificial Intelligence & Daily Life (credits $2$) (overview for non-specialists). ### Semester II – Core Highlights #### CAI-CC-521 AI Systems Engineering - OOP vs procedural, UML (use-case, class, activity, sequence, state), agile & scrum, story-boarding. - MLOps – architecture, pipelines, risk, tool-chain, real-world cases. #### CAI-CC-522 Database Systems for Big Data - $3V$ big-data traits; Hadoop (HDFS, YARN), Spark (RDD), NoSQL (MongoDB CRUD), CAP theorem, file formats (Avro/Parquet), Uber & Google case studies. #### CAI-CC-523 Machine Vision & Pattern Recognition - Human vs machine vision; image processing (filters, FFT), segmentation (watershed), features (LBP, GLCM, SIFT/SURF), distance metrics (Euclidean, Bhattacharyya), max-likelihood, kernel density, Fisher LDA, applications (YOLO, CBIR). #### CAI-CC-524 Pattern Recognition Lab - Cycles: image pre-processing, computer vision, unsupervised ML, supervised ML; evaluates via projects + viva. ### Semester II – Department Electives (samples) - Blockchain Technology: consensus (PoW, PoS), Bitcoin internals, Ethereum & Solidity, SNARKs, big-data use cases. - IoE: architecture, M2M, cloud + AI for smart systems. - Cyber-Security & Cyber-Law: threats, risk, forensics, IT Act $2000$, GDPR-style privacy. - Statistical ML Techniques: convergence, regression (Lasso/Ridge), K-NN, EM, Markov/HMM, Bayesian nets, real-world DNA example. - Data Mining & Text Analytics: preprocessing, OLAP, advanced frequent-pattern mining, VSM, TF-IDF, word2vec. ### Semester III – Core Courses #### CAI-CC-531 Computational Cognitive Systems - Tri-level hypothesis; cognitive/neuro/network/Linguistic/AI perspectives. - Vision & pattern recognition theories (Template, RBC). - Memory & problem-solving models (ACT*, SOAR). - Neuroscience tools (CT, PET, fMRI). - Semantic & neural networks; natural language processing pipeline; AI methodologies & agent design. #### CAI-CC-532 Applied Machine Learning - PCA math (eigenvectors), visualisation, fuzzy/ANFIS, ensemble (AdaBoost, XGBoost, stacking), associative memories (Hopfield, BAM), SOM, ART, PNN, optimisers (Adam), real apps (malware, HAR, recommender). #### CAI-CC-533 Accelerated NLP - Pipeline: tokenise → stem/lemma → POS/NER. - HMM/MEMM tagging. Vector space (TF-IDF). - Word semantics (WordNet, word2vec, GloVe). - Deep NLP: RNN/LSTM, attention, Transformer; BERT, GPT. - Tasks: summarisation, embedding comparison, document classification. #### CAI-CC-534 Machine Interaction Lab - Raspberry-Pi starter: install, GPIO, LED, LCD, web-server. - Sensor + ML cycle: real-time object/face recognition, OCR label reading, licence-plate. - Assessed via $30+50+20$ marks. #### CAI-CC-535 Case Study (credits $2$) - Choose peer-reviewed article → replicate → write technical report + present (20 interaction + 30 report + 50 viva). ### Semester III – Department Electives (selection) - Foundation in Robotics: kinematics (D-H), sensors, grippers, programming & specs. - Game Theory & Applications: Nash equilibrium, Bayesian games, mechanism design (VCG, auctions). - Speech Processing & Recognition: STFT, MFCC, HMM, ASR metrics. - Nature-Inspired Computing: GA, ACO, PSO, ABC, SA, Firefly, Cuckoo, bat, applications. - Intelligent Information Retrieval: Boolean/vector/probabilistic models, ranking (PageRank, HITS), clustering, recommender TF-IDF. - Special Topics CAI-DE-537 series - AI Planning (STRIPS, HTN, MDP, $A^*$, uncertain planning). - Methods for Causal Inference (randomised vs observational, DAGs, do-calculus, propensity, doubly-robust). - Deep Architectures (CNN, RNN, LSTM, GAN, normalization). - Computational Creativity (creative algorithms, evaluation, AI art/music). - Evaluation of AI Systems (precision/recall, ROC, bias-variance, drift, explainability LIME/CAM, test risks). ### Skill / Generic Components (Sem III)