Artificial Intelligence · Unit 5 · 9 hrs
Machine Learning
Exam-focused notes for Machine Learning (Artificial Intelligence, CSC266): what the TU syllabus asks and how it has actually been tested, with 12 solved past questions from this unit.
What this unit covers
- Introduction to Machine Learning
- Concepts of Learning
- Supervised, Unsupervised and Reinforcement Learning
- Statistical-based Learning: Naive Bayes Model
- Learning by Genetic Algorithm
- Learning with Neural Networks: Introduction, Biological Neural Networks Vs. Artificial Neural Networks (ANN), Mathematical Model of ANN, Types of ANN: Feed-forward, Recurrent, Single Layered, Multi-Layered, Application of Artificial Neural Networks, Learning by Training ANN, Supervised vs. Unsupervised Learning, Hebbian Learning, Perceptron Learning, Back-propagation Learning
Learning with Neural Networks
How can you relate synapse, dendrite, and axon in biological neural networks with the elements of artificial neural networks? Create a multi-layer ANN with input layer, hidden layer, and output layer. Assume necessary inputs and weights to the ANN and illustrate a single iteration of backpropagation algorithm to train the ANN.[10]
The question requires us to assume inputs and weights, so the worked example below uses these values: Inputs: $x1 = 0.5$, $x2 = 0.3$ Target: $t = 1.0$ Learning rate: $\alpha = 0.5$ Activation: Sigmoid $g(x) = \dfrac{1}{1+e^{-x}}$, derivative $g'(x) = g(x)(1...
Full solved answer →What is reinforcement learning? Configure an ANN neuron to simulate OR gate. [5]
This is a conceptual + design question. The relevant "data" is: - Task 1: Define reinforcement learning. - Task 2: Configure a single ANN neuron (perceptron) to implement the OR logic function. - OR gate truth table (standard, not given but implied): $x1$ $...
Full solved answer →What is the role of activation function in ANN. How sigmoid function works? Discuss about perceptron learning.[10]
--- An activation function determines the output of a neuron given its input. It introduces non-linearity into the network, enabling ANN to learn complex patterns. Role Description ------------------- Non-linearity Allows the network to learn non-linear map...
Full solved answer →Give an example of reinforcement learning. Explain the types of ANN. [5]
--- Reinforcement Learning is a type of dynamic learning that trains an agent using reward and punishment. The agent learns by interacting with its environment. It consists of three components: - Agent (the learner) - Environment (what the agent interacts w...
Full solved answer →Describe mathematical model of neural network. What does it means to train a neural network? Write algorithm for preceptron learning.[10]
--- A neural network is inspired by the biological neuron. The mathematical model of a single artificial neuron (also called a perceptron or node) consists of the following components: Component Description ------------------------ Inputs x₁, x₂, ..., xₙ (f...
Full solved answer →Define mathematical model of artificial neural network. Discuss how Hebbian learning algorithm can be used to train a neural network. Support your answer with an example.[10]
An Artificial Neural Network (ANN) is inspired by the biological neural network of the human brain. The mathematical model of a single artificial neuron (also called a McCulloch-Pitts neuron, first proposed in 1943) consists of the following components: Sym...
Full solved answer →Supervised, Unsupervised and Reinforcement Learning
What is reinforcement learning? Configure an ANN neuron to simulate OR gate. [5]
- Task 1: Define reinforcement learning (conceptual). - Task 2: Configure a single ANN neuron (perceptron) to simulate the OR gate. - OR gate truth table (standard, boolean logic): $x1$ $x2$ $y = x1 \lor x2$ ---------------------------------- 0 0 0 0 1 1 1 ...
Full solved answer →Differentiate supervised learning from unsupervised? Discuss how Naive Bayes Model can be used for machine learning? Support your answer with example.[10]
--- The system is supplied with a set of training examples consisting of inputs and corresponding outputs (labelled data). It can take what it has learnt in the past and apply that to new data using labelled examples to predict future patterns and events. C...
Full solved answer →Learning by Genetic Algorithm
Define selection, crossover, and mutation operations in genetic algorithm. [5]
--- Selection is the operator that determines which individuals from the current population are chosen as parents to produce the next generation. - Individuals with higher fitness scores are given preference and are more likely to pass their genes to the ne...
Full solved answer →What is iscrossver operation in genetic algorithm? Given following chromosomes show the result of one-point and two point crossover. C1 = 01100010, C2 = 10101100. Choose appropriate crossover points as per your own suggestions. [5]
- Chromosome $C1 = 01100010$ - Chromosome $C2 = 10101100$ - Each chromosome has 8 bits. - Crossover points to be chosen by student. Position indexing: --- Crossover (also called recombination) is a genetic operator in a Genetic Algorithm that combines the g...
Full solved answer →Write an algorithm for learning by Genetic Approach. [5]
Genetic Algorithms (GAs) are adaptive heuristic search algorithms that belong to the larger part of evolutionary algorithms. They are based on the idea of natural selection and genetics. Historical data are provided to find better solutions; GAs are used to...
Full solved answer →Discuss how genetic algorithm works? [5]
Genetic Algorithms (GAs) are adaptive heuristic search algorithms that belong to the larger part of evolutionary algorithms. They are based on the idea of natural selection and genetics. Historical data are provided to find better solutions, or simply GAs a...
Full solved answer →Make Unit 5 stick
Practice CSC266 with flashcards & quizzes