handshakeExam Gist

All Important topic, related to exam

ML

Exam will be on paper only, even for MCQ ans

chevron-rightBasic Q/N for examhashtag
  • Duration: 3 hours

  • Mode: Question paper will on LMS and upload their only

  • Exam Type: Closed book

  • Coding: Not required

  • Derivations: Only one solved derivation may be asked

  • Focus: Concept-driven problems

  • Answer Format: No lengthy answers required

  • Device: Paper only (no iPad)

  • Answer Submission: Write all answers (MCQ and subjective) on paper, not on LMS

  • Calculator: Scientific calculators allow

chevron-rightTopicshashtag
  1. ML Definitions

  2. Data Pre-processing

    • Performance Measures

    • Training and Testing

    • Cross Validation

    • Correlation

    • Imputation

    • Redundant Data Removal

  3. Linear Regression

    • Regression Terminology

    • Gradient Descent & Types

    • Ridge & Lasso Regression

    • Normal Equation

    • Regularization

    • Bias-Variance Tradeoff

    • Locally Weighted Regression (LWR)

    • Derivation

  4. Classification

    • Logistic Regression

    • Multiclass Classifier

    • Cost Function

    • Accuracy, Precision, Recall, TP, TN

    • ROC Curve (Evaluation Matrix)

    • Confusion Matrix

  5. Bayesian Classifier

    • Bayes Theorem

    • Classifier Terminology

    • Naive Bayes Theorem

    • Covariance Matrix

    • Univariate / Multivariate Densities

    • Risk Function

  6. Decision Tree

    • Structure & Terminology (Root, Leaf, etc.)

    • Tree Design

    • Impurity Measures: Gini, Entropy

    • CART, ID3 (Multiclass)

    • Information Gain

    • Decision Tree Regularization

  7. Random Forest

    • Ensemble Methods

    • Voting Classifier

    • Bagging

    • Random Subspace Method

    • OOB Instances

    • Boosting

    • Regression-based Random Forest

  8. SVM (Support Vector Machine)

    • Terminology

    • Regularization

    • Parameters & Hyperparameters

    • Cost Function

    • Non-linear SVM

    • Kernel Techniques

    • Derivation of Optimal Margin

  9. K-Means

    • Unsupervised Learning

    • Applications

    • Algorithm (with Pseudocode)

    • Cost Function

    • Elbow Method / Silhouette Score

  10. PCA (Principal Component Analysis)

    • Eigenvectors & Eigenvalues

    • Eigen Decomposition

    • SVD & Types of SVD

    • Variance & Covariance

    • U, Sigma Matrix

    • Projection & Manifold Learning

    • Eigenface Problem

  11. Neural Networks

    • Importance

    • Perceptron, LTU, TLU

    • Weights & Bias

    • Notation & Matrix Formulation

    • Forward Pass

    • Activation Functions

    • Gradient Descent

    • Backpropagation

    • Regularization

  12. CNN (Convolutional Neural Network)

    • Architecture

    • Purpose

    • Kernels, Striding, Padding

    • Pooling

    • Example Architectures

    • Backpropagation in CNN

  13. Autoencoders

    • Architecture

    • Loss Function

    • Variational Autoencoders (VAE)

    • Generative Models

  14. RNN (Recurrent Neural Network)

    • Structure

chevron-rightNumerical problemshashtag
  • Information Gain

  • Design a Spam Filter using Bayesian Classification

  • Neural Network Backpropagation

  • Confusion Matrices

  • Boosting

  • Equation of a Regression Line (given data points)

  • Distance from Hyperplane – SVM

  • Logistic Regression

  • Data Preprocessing Techniques

  • CNN Concepts – Striding, etc.

file-pdf
386KB
file-pdf
80KB
file-pdf
795KB

DSA

Paper will be MCQ only

chevron-rightTopicshashtag
  • After Minor (weightage 70%)

    • Red-Black Trees

    • Red-Black Tree Deletion

    • Augmenting Data Structures

    • Disjoint-Set

    • Disjoint-Set as Trees

    • Graph Algorithms

      • Breadth-First Search (BFS)

      • Dijkstra’s Algorithm

    • Flow Networks

      • Ford-Fulkerson Method

      • Bipartite Maximum Matching

    • Dynamic Programming

      • Longest Common Subsequence (LCS)

    • Complexity Classes

      • P, NP, and NP-Completeness

  • Before Minor (weightage 30%)

    • About Algo

    • Insertion Sort

    • Merge Sort

    • Recurrence Relation

    • Hashing

    • Binary Search Tree

    • Binary Search Tree Deletion

ODS

chevron-rightTopicshashtag
  • Fact-1 NLLP (Nonlinear Linear Programming)

    • Theory

      • Graphical Method

      • Vector and Tuple, Vector Space

      • Convex Set, Convex Combination

      • Convex Function (∪ shaped), Concave Function (∩ shaped)

      • Hessian Matrix

        • Determinant calculations

        • Symmetric Matrix

        • Positive/Negative Definite Hessian (Local Minima/Maxima)

        • Convexity/Concavity via Hessian

      • Global Minima/Maxima

      • Orthogonal Vectors

    • Method

      • First & Second Order Conditions

      • Line Search Method

        • Unimodal Function

        • Golden Section Method

        • Fibonacci Search Method

      • Decent method

        • Taylor Series Expansion

        • Newton's Method

          • Single Variable Newton Method

          • Multivariable Extension

        • Steepest Descent Method

          • Directional Derivatives

        • Quasi Newton Method

      • Conjugate Gradient Method

        • Conjugate Directions

      • Least Square Approximations

  • Fact-2

    • Graphical Method

    • Simplex & Big M Method

    • Lagrange Multipliers & KKT Conditions

    • Penalty and Barrier Methods

    • Frank-Wolfe Algorithm

chevron-rightPaternhashtag
  • Fact-1

    • 4 Subjective Questions (guess)

    • Remaining Objective Questions

  • Fact-2

    • 4 Subjective Questions

    • Remaining Objective Questions

file-pdf
61KB

AI

chevron-rightTopicshashtag
  • Basic AI

  • AI Agent

  • Solving Problems by Searching

  • Knowledge-Based Agents

  • First Order Logic

  • Local Search

  • Constraint Satisfaction Problems

  • Adversarial Search and Games

  • Markov Decision Process and Reinforcement Learning

  • Bayesian Networks

  • Automated Planning

chevron-rightPaternhashtag
  • Question Type:

    • 3 Subjective Questions

    • Remaining Objective Questions

  • Difficulty Level:

    • 25% Easy

    • 50% Medium

    • 25% Difficult

file-pdf
701KB

Last updated