Probability for GATE Machine Learning
Probability is the grammar of ML: classifier scores, Naive Bayes tables and confidence statements are conditional probabilities. GATE tests whether you condition correctly and update beliefs with evidence.
Important concepts
- Conditional probability P(A|B) and Bayes theorem.
- Independence vs conditional independence (Naive Bayes).
- Expectation, variance, linearity of expectation.
- Bernoulli, binomial, Gaussian, exponential distributions.
- Law of total probability as the Bayes denominator.
GATE relevance
Direct Bayes arithmetic, Naive Bayes tables, P(A|B) vs P(B|A) traps, and likelihood vs posterior framing inside ML questions.
Important formulas
P(A|B)=P(A∩B)/P(B) · P(C|E)=P(E|C)P(C)/P(E) · E[ΣXi]=ΣE[Xi]
Worked example
Test 90% sensitive, 95% specific, 1% sick: P(sick|pos)=0.9·0.01/(0.9·0.01+0.05·0.99)≈15%. Base rates dominate — the exact reasoning Naive Bayes mechanises.
Practice questions
Run GATE ML practice questions, then a timed mock. Theory: Guide Unit 1 + maths refresher.
Related topics
Statistics · Linear algebra · Machine learning
FAQ
Which distributions matter most?
Bernoulli, binomial, Gaussian, exponential — parameters, mean, variance, and where each appears.
Common mistake?
Confusing P(A|B) with P(B|A) and dropping the base rate.