MCA-203
Artificial Intelligence and Machine Learning
Classical AI and modern machine learning in one paper — search and knowledge representation, then regression, classification, clustering and neural networks.
- Semester
- Semester 3
- Credits
- 3
- Scheme
- 3 lectures a week
- Contact hours
- 40 across four units
- Category
- Core
What the paper is
Artificial Intelligence and Machine Learning is really two subjects sharing a paper, and the join is visible in the middle. The first half is classical AI out of Russell and Norvig: state space search, heuristics, and logic-based knowledge representation and reasoning. The second half is statistical learning out of Hastie: regression, classification, clustering, ensembles and neural networks.
The consequence for revision is that the two halves are examined in completely different registers. Units I and II want you to trace A* on a graph, or resolve a set of predicate clauses. Units III and IV want you to compute a confusion matrix, explain the bias-variance tradeoff, or work through one step of backpropagation.
The syllabus also names ethics of AI in Unit I, which is easy to skip as soft material and is explicitly listed. It is a small number of marks that costs almost nothing to secure.
What to concentrate on
- The search algorithms are the most reliably examinable part of Unit I. Be able to run A* and AO* on a given graph with a given heuristic and say why a heuristic is admissible — that trio covers most of what gets asked.
- For Unit II, the inference rules are a closed list — modus ponens, modus tollens, resolution, and-elimination, syllogism. Resolution in predicate logic is the one worth practising, because it is procedural and it is where the longer question usually goes.
- In the machine learning half, the accuracy measures are the connective tissue: confusion matrix, precision, recall, the coefficient of determination, and how overfitting shows up in each. Nearly every Unit III and IV question touches one of them.
Syllabus, unit by unit
Summarised from the approved scheme. Hours are the contact hours the scheme allots, which is a fair guide to where the marks sit.
Unit I · AI fundamentals and search
10 hours- Defining artificial intelligence, types of AI, the Turing test, AI techniques
- Comparison of AI, machine learning and deep learning
- Problem characterisation and reduction, defining state space
- Search techniques: hill climbing, breadth first and depth first search, best first search, A*, AO*, constraint satisfaction, means-ends analysis
- Ethics of AI
Unit II · Knowledge representation and reasoning
10 hours- Approaches to knowledge representation: propositional logic, first-order predicate logic
- Inference rules: modus ponens, modus tollens, resolution, and-elimination, syllogism
- Production rules and types of knowledge
- Forward and backward reasoning, non-monotonic reasoning, reasoning with uncertainty
Unit III · Supervised learning
10 hours- Introduction to machine learning, types of problems and applications
- Regression: simple and multiple regression, least squares, total sum of squares, sum of squares of residuals and of regression, odds and odds ratio
- Classification: logistic regression
- Accuracy methods: coefficient of determination, correlation, confusion matrix
- Overfitting and underfitting, bias and variance
Unit IV · Unsupervised learning and neural networks
10 hours- Dimensionality reduction with principal component analysis
- K-means clustering
- Ensemble learning: boosting and bagging
- Neural networks: types, activation functions, feed forward, the backpropagation algorithm
- Recommender systems and content-based recommendations
How it is examined
- Question 1 is compulsory, covers the whole syllabus, and is made of objective or short-answer questions. It carries 12 marks.
- The rest of the paper is four units matching the four above. Each unit sets two questions and you attempt one, so there are four more answers at 12 marks each.
- Examiners are asked to map every question to a course outcome using Bloom's taxonomy, which is why questions are phrased as “apply”, “analyse” or “evaluate” rather than only “explain”.
The pass mark is 40 out of 100 in every paper.
Learning objectives
- AI principles and approaches
- Understanding the building blocks of AI as intelligent agents: search, knowledge representation, inference, logic and learning
- Understanding the nature of problems solved with machine learning
Assumed beforehand
- Basic knowledge of mathematical logic
- Linear algebra
Paired in the scheme with MCA-263 Artificial Intelligence and Machine Learning Lab.
Books it is taught from
- TB1Stuart Russell and Peter Norvig, “Artificial Intelligence: A Modern Approach”, Pearson Education, 4th edition, 2020
- TB2T. Hastie, R. Tibshirani and J. Friedman, “The Elements of Statistical Learning”, Springer, 2nd edition, 2017
- TB3Richard O. Duda, Peter E. Hart and David G. Stork, “Pattern Classification”, Wiley, 2nd edition, 2012
Notes for MCA-203
Lecture notes, question papers and assignments filed against MCA-203 are in the library. Everything there has been checked by a moderator before it appears, and an account is free — new ones are approved by hand, which is what keeps the shelf clean.