Chapter 2 - Supervised Network (Perceptron)

Question 1

What are Artificial Neural Networks (ANNs) described as?

Answer 1

hey are described as self-learning mechanisms that do not require traditional programming skills

Question 2

What is a key component of an artificial neural network?

Answer 2

It consists of interconnected processors called neurons

Question 3

How do neurons in an ANN communicate?

Answer 3

They communicate through weighted links that pass signals from one neuron to another

Question 4

How many output signals does a neuron produce?

Answer 4

A neuron produces only one output signal

Question 5

What is the role of weights in ANNs?

Answer 5

Weights are a means of long-term memory, and they express the strength or importance of connections

Question 6

How does a neural network learn?

Answer 6

It learns by repeatedly adjusting the weights

Question 7

What is the main idea behind supervised learning?

Answer 7

Training a model on a labeled dataset, where input data is paired with the correct output, allowing the model to learn how to map inputs to outputs

Question 8

How does a supervised learning model learn?

Answer 8

By minimizing the error between its predictions and the actual labels

Question 9

What are the two main phases in supervised learning?

Answer 9

Training and testing

Question 10

Give an example of a supervised learning task.

Answer 10

Image classification, spam detection, predictive maintenance, and medical diagnosis are all examples of supervised learning tasks

Question 11

What is the goal of regression?

Answer 11

To predict a continuous outcome based on one or more input features

Question 12

What is linear regression?

Answer 12

The simplest form of regression, which models the relationship between the input and output as a straight line

Question 13

How is linear regression visualized?

Answer 13

Using scatter plots with a regression line and plots showing the cost function’s surface with gradient descent

Question 14

Why is linear regression a good starting point for machine learning problems?

Answer 14

Because it is simple and interpretable, especially when the relationship between the features and the target is linear

Question 15

What are the limitations of linear regression?

Answer 15

It is limited when the relationship between input and output is nonlinear, or when there are complex interactions between features

Question 16

What is a neuron described as in the context of computing?

Answer 16

A simple computing element that receives input signals, computes an activation level, and sends an output signa

Question 17

What is the purpose of an activation function?

Answer 17

To introduce non-linearity into the output of a neuron and decide whether it should be activated

Question 18

What is a binary step function?

Answer 18

A threshold-based activation function where the neuron is activated if the input is above or below a certain threshold

Question 19

How does the sign activation function work?

Answer 19

It is a threshold-based activation function that produces +1 or -1 based on input

Question 20

What are the characteristics of the Sigmoid/Logistic activation function?

Answer 20

It produces a smooth gradient, bounds output values between 0 and 1, and enables clear predictions by bringing the output close to 1 or 0 when the input is above 2 or below -2

Question 21

How does a perceptron produce an output?

Answer 21

The output is generated by applying an activation function to the weighted sum of inputs plus a bias

Question 22

What are the steps involved in training a perceptron?

Answer 22

Initialization, activation, weight training/adjustment, and iteration

Question 23

How is the error calculated in a perceptron?

Answer 23

Error is calculated as the difference between the desired output and the actual output

Question 24

How are the weights updated in a perceptron?

Answer 24

The weight change is calculated by multiplying the learning rate (alpha), the input, and the error; the current weight is then updated by adding the weight change to it

Question 25

When does the training iteration of a perceptron end?

Answer 25

When the perceptron reaches zero error for all possible input-output pairs

Question 26

Can a perceptron learn the XOR operation?

Answer 26

The source suggests this as an exercise without providing the answer, but it implies that a simple perceptron cannot learn XOR

Question 27

What is the definition of a linearly separable function in a 3-dimensional space?

Answer 27

The source prompts the reader to think about the concept without defining it explicitly