Skip to main content
For any enquiries please WhatsApp at +91 7893756364
GATE DA Preparation7 min read

Artificial Intelligence GATE DA: Syllabus & Books

AI for GATE DA 2027: search (BFS/DFS/A*), propositional & first-order logic, Bayesian nets — Russell & Norvig + Rosen logic plan.

29 May 2026

AI on GATE DA splits into three clean pillars — search, logic, and reasoning under uncertainty — and each pillar has its own best book. Russell & Norvig handles search and probabilistic reasoning. Kenneth Rosen handles logic far better than R&N for exam-style problems. The third pillar (Bayesian networks, Markov models) is the one aspirants skip most often and regret most on exam day. If you treat AI as "just search algorithms", you are leaving a third of the subject unprepared.

Verify: Confirm the syllabus on the official GATE 2026 syllabus page (IIT Guwahati).

Official Syllabus

  • Search — informed, uninformed, adversarial
  • Logic — propositional, predicate
  • Reasoning under uncertainty — conditional independence representation, exact inference through variable elimination, and approximate inference through sampling

The three pillars do not overlap heavily with each other (except the probability link in pillar three), so phased study works naturally. Note: the official syllabus uses "predicate" logic (equivalent to first-order logic / FOL). Bayesian networks and Markov models are standard frameworks for the reasoning-under-uncertainty topics listed above.

The Three-Pillar Map

Pillar Topics Best book Overlap
SearchBFS, DFS, UCS, greedy best-first, A*; completeness/optimality/complexityRussell & NorvigDSA — BFS, DFS, shortest path
LogicPropositional logic, predicate (first-order) logic, inference, CNF/DNF, resolutionKenneth RosenDiscrete Mathematics (logic chapters)Formal methods
Reasoning under uncertaintyConditional independence, exact inference (variable elimination), approximate inference (sampling)Russell & NorvigProbability, ML (naive Bayes)

Books and Resources

Resource Role Use for Skip
Russell & NorvigAI: A Modern ApproachPrimary (search + reasoning)Search algorithms, Bayesian networks, Markov modelsRobotics, perception, NLP chapters
Kenneth RosenDiscrete Mathematics and Its ApplicationsPrimary (logic)Propositional and FOL, inference, normal formsCombinatorics beyond GATE scope
Official GATE DA PYQs (2024 onwards)PracticeSearch traces, logic conversion, Bayes-net inference

Full book list: GATE DA books and resources guide.

Search — Pillar 1

Uninformed Search

BFS, DFS, depth-limited, iterative deepening, uniform-cost search. Be fluent on completeness, optimality, time and space complexity for each — common conceptual MSQ targets.

Informed Search

Greedy best-first uses only h(n). A* uses f(n) = g(n) + h(n). A* is complete and optimal when h is admissible (never overestimates true cost); with a consistent heuristic, A* is also optimal in the graph-search setting.

Algorithm Comparison Card

Print this. It is the single most useful reference for search PYQs.

Algorithm Complete? Optimal? Time Space
BFSYes (finite branching)Yes (unit costs)O(bd)O(bd)
DFSNo (infinite spaces)NoO(bm)O(bm)
UCSYes (non-negative costs)YesO(b1+⌊C*/ε⌋)O(b1+⌊C*/ε⌋)
GreedyNoNoO(bm)O(bm)
A*Yes (admissible h)Yes (admissible h)Exponential worst-caseExponential worst-case

Logic — Pillar 2 (Use Rosen)

Propositional Logic

Atomic propositions, connectives (¬, ∧, ∨, →, ↔), truth tables, tautologies, contradictions, satisfiability. Equivalence laws (De Morgan, distribution, contrapositive). CNF and DNF — converting an arbitrary formula to CNF is a common PYQ.

First-Order Logic

Quantifiers (∀, ∃), predicates, variables, constants, functions. Translating English ↔ FOL is the standard PYQ pattern. Be fluent on negation with quantifiers: ¬∀x P(x) ≡ ∃x ¬P(x).

Inference

Modus ponens, modus tollens, hypothetical syllogism, resolution. Resolution refutation: convert to CNF, negate the goal, derive the empty clause. For FOL, add unification.

Why Rosen over R&N here: Rosen's exposition of inference rules and CNF/DNF conversion reads more like an exam-style treatment. R&N's logic chapters lean toward AI-system implementation. Either is correct; Rosen is faster for prep.

Reasoning Under Uncertainty — Pillar 3

Bayesian Networks

DAG where nodes = random variables, edges = conditional dependence, each node has a CPT given its parents. D-separation lets you read conditional independence from the graph structure.

Inference: enumeration (exact, exponential), variable elimination (exact, more efficient), approximate inference through sampling (rejection, likelihood weighting, Gibbs). The prerequisite is Bayes' theorem from probability; naive Bayes from the ML syllabus is essentially a Bayes net with fixed structure.

Note on Markov models and MDPs: The official GATE DA syllabus does not explicitly list Markov chains, HMMs, or Markov decision processes under AI. However, Markov models are a natural extension of the conditional-independence framework the syllabus names, and some coaching material covers them. Focus your time on the explicitly listed topics (conditional independence representation, variable elimination, sampling-based inference) and treat Markov models as supplementary awareness only.

Past-Paper Patterns

AI PYQs since 2024 cluster into six types:

  1. Trace BFS / DFS / UCS / A* on a small graph
  2. Identify completeness / optimality / complexity from algorithm definitions
  3. Translate English ↔ FOL or convert propositional formulas to CNF
  4. Resolution refutation
  5. Inference in a small Bayesian network
  6. Conditional independence from a Bayes-net structure

Solve every AI PYQ from GATE DA 2024 and 2025. Graph search overlaps with DSA — reuse that preparation.

What to Skip

  • Robotics — motion planning, kinematics
  • Perception — computer vision beyond conceptual exposure
  • NLP — parsing, translation, LLMs (the R&N NLP chapters)
  • Adversarial search in depth — minimax/alpha-beta useful conceptually, rarely tested at depth
  • Reinforcement learning — MDPs, Q-learning, policy gradients (not in official syllabus)
  • CSPs beyond basic exposure

Study Sequence

Weeks 1–2: Search

  1. Uninformed search — BFS, DFS, UCS, iterative deepening; properties and complexities.
  2. Informed search — greedy, A*; admissibility and consistency.
  3. Trace practice on small graphs.
  4. Build the comparison card (table above).

Weeks 3–4: Logic (Rosen)

  1. Propositional logic — connectives, truth tables, equivalences, CNF/DNF.
  2. FOL — quantifiers, English ↔ FOL translation.
  3. Inference — modus ponens/tollens, resolution.

Weeks 5–6: Reasoning Under Uncertainty

  1. Bayesian networks — structure, d-separation, inference.
  2. Supplementary awareness of Markov models (not explicitly in syllabus).

Week 7: PYQs and Revision

  1. Every AI PYQ from GATE DA 2024 and 2025.
  2. 2–3 topic-wise tests from The ML Hub GATE DA test series.
  3. Finalise comparison card, logic-equivalence sheet, Bayes-net inference template.

Traps to Avoid

  • Using R&N for everything including logic. Rosen is faster for exam prep; use each book in its strength zone.
  • Skipping reasoning under uncertainty. Bayesian networks are explicitly in the syllabus and the least-prepared pillar.
  • Memorising search complexities without trace practice. PYQs test behaviour, not memorised tables.
  • Confusing admissibility and consistency. Admissibility: h never overestimates. Consistency: adds triangle inequality.
  • Missing the DSA overlap. Graph traversal can appear under AI or DSA.
  • Over-studying Markov models / MDPs. Not explicitly in the official syllabus; keep to supplementary awareness.

Find your weak pillar

The AI module in our free GATE DA demo course includes lectures on search, logic, and Bayesian networks plus a benchmark test — useful for identifying which pillar needs the most work.

AI in The ML Hub's Course

The AI block in The ML Hub's GATE DA course follows the three-pillar map. Search uses the comparison table directly; logic uses Rosen as the primary reference; reasoning under uncertainty connects to the probability prerequisite. Topic-wise tests include search trace, FOL conversion, and Bayes-net inference problems.

Three pillars. Two books. One structured programme.

AI is structurally clean — but only if you respect the pillar split and use the right book for each.

  • Mentor-led lectures: search (R&N), logic (Rosen), Bayesian networks (R&N)
  • Search comparison card and Bayes-net inference templates from GATE DA rankers
  • Topic-wise tests on search, logic, and reasoning in the test series

Explore the Course · Test Series · Schedule · Free Demo

FAQs

Which book for GATE DA AI?

Russell & Norvig for search and reasoning under uncertainty. Kenneth Rosen's Discrete Mathematics (logic chapters) for propositional and first-order logic — Rosen's exposition is clearer for exam-style problems.

Is first-order logic in GATE DA?

Yes. Cover quantifiers, predicates, English ↔ FOL translation, negation with quantifiers, and resolution-based inference.

Are Bayesian networks in the GATE DA syllabus?

The official syllabus lists conditional independence representation, exact inference through variable elimination, and approximate inference through sampling — Bayesian networks are the standard framework for these topics. Cover graph structure, d-separation, variable elimination, and sampling-based inference.

Is A* search in GATE DA?

Yes. Cover f(n) = g(n) + h(n), admissibility, consistency, and the resulting completeness/optimality guarantees.

Related Guides

Subjects that share machinery with AI: Python & DSA (graph search overlap) and Machine Learning (naive Bayes, probabilistic reasoning). Full subject map: GATE DA books and resources · GATE DA syllabus 2027 guide.

Related topics
gate-daartificial-intelligencegate-da-2027preparationrussell-norvigsearch-algorithmsbayesian-networksfirst-order-logic
Start Preparing Today

Ready to Crack GATE DA?

Join The ML Hub for structured courses, daily practice, mock tests, and 1:1 mentorship from GATE toppers and IIT alumni.