Neural Networks | CBSE Class 10 AI Project Cycle Notes – 6

Neural Networks | CBSE Class 10 AI Project Cycle Notes – 6

1. What do you mean by a neural network in the context of computing?
OR
What do you mean by an Artificial Neural Network (ANN)?

It is defined as collection of connected computational components called neurons to process information just like human brain.

2. Explain different layers of ANN.

There are three layers of artificial neural network.
i. Input Layer: This layer is used to feed data to the neural network. It performs no processing.
ii. Hidden Layer: It is interconnected collection of nodes that perform different functions in a neural network. It receives input from input layer, does required processing and provides output to output layer. The processing carried out at hidden layer is as:
Sum of weighted inputs + activation function + hidden rules
iii. Output Layer: This layer receives information from hidden layer and gives final output.

3. How does a neural network work?

In a neural network, information gets transferred from input layer to hidden layer for processing and decision making and then final result is transferred to output layer.
Neural networks are initially trained with some data and they keep learning with every new input.

4. How are artificial neural networks different from normal computers?

Artificial neural networks learn through training data but they need a long time to learn and produce satisfactory results.
Normal computers perform different tasks due to programs stored in them and they start producing results immediately after it gets programmed.

5. What do you mean by a neuron in ANN?

It is also known as a node. It can be defined as one computational unit in artificial neural network. It can perform one specific task.

6. How does an artificial neuron work?

Artificial neuron receives values through input and then depending upon its algorithm and weight of input signal, it generates an output which is received by other neurons.

7. How does training of artificial neural network happen?
OR
What happens at the artificial neuron when training of artificial neural network takes place?

  1. Initialize weights of all neurons.
  2. Input layer should get one set of input data (exemplar).
  3. Calculate the outputs as per the weights and activation function.
  4. Compare the output with expected results.
  5. Update weights if the output doesn’t match with expected result.
  6. Repeat above steps for multiple sets of input data.

8. What is the role of an activation function in a neural network?

It defines how the weighted sum of inputs is transformed into an output of a node.

9. How do neural networks learn and improve themselves?

When we provide an input to an artificial neural network, it modifies it to generate an output. On the basis of difference between the output and actual result, the neurons in ANN make changes in the weights to produce the correct output.

10. Write features of a Neural network.

  1. They can mimic the working of human brain.
  2. They can automatically learn with each input.
  3. They can work with big datasets.
  4. They use machine learning techniques to function and evolve.

11. Write advantages of ANNs

  1. They can learn and perform multiple tasks parallelly.
  2. They can work even any of its part doesn’t function.
  3. They can learn and generalize.

12. Write disadvantages of ANNs

  1. They need very large amount of diverse data for training.
  2. They are unable to tell the reason behind a particular result they produce.
  3. They are computationally more expensive.
  4. Their development takes very long time.

13. What is back propagation? What is its significance in the training of ANNs?

It is the method to train an artificial neural network. It can reduce the difference between the model’s predicted output and the actual output by adjusting the weights and biases in the network. It helps to make suitable changes to produce correct output.

14. What are the two main abilities of neural networks?

i. Pattern recognition: Neural network can identifying patterns within data, recognize speech and images etc.
ii. Prediction and classification: Neural network can predict the results in which classification of data done.

15. What do you understand by the black box nature of ANN?

Neural networks are considered as black box in nature because their decision making processes are:
i. Opaque: Difficult to understand and interpret
ii. Complex: Interactions between layers and neurons are very complex.
iii. Non- intuitive: They can display unpredictable behaviour.

16. What do you mean by overfitting?

It is a model’s excessive adaptation to training data. It gives high accuracy on training data but low accuracy on new data.

Q17. Discuss some common applications of neural networks.

Character recognition: It is the process of recognizing text inside images and converting it into an electronic form. This images could be handwritten text, printed text like document , Receipt etc.
Speech recognition : Artificial neural networks can recognize speech and give response accordinguy. For example: Alexa, Siri etc.
Computer vision: With the help of NNs, computer can accurately process and understand visual data.
Stock market Prediction: Neural networks can examine a lot of information quickly. so they help in stock market prediction.
Health care: Neural network helps in medical image analysis, disease diagnosis, personalized medicine etc.
Autonomous systems: Neural network are used in self driving cars, drones, robotics, smart homes etc.
Recommendation systems: Product recommendation, movie recommendation, job recommendation, travel recommendation etc.

18. Define the following:

Neuron :- A neuron consists of weighted inputs (dendrites), an activation function(soma) and one output(axon).
Weighted input:– It is a numerical value associated with the connections between neurons across different layers of the network. It signifies the strength and direction of the influence of one neuron over another.
Axon:- It is the output produced by a neuron.
Synapse:-It is the connection from a neuron to another that carry the information.
Activation function:-It defines how the weighted sum of the input is changed into an output.
Bias:– It refers to the errors in the learning process. It can result into inaccurate outputs and predictions.

Related Links

Spread the love
Lesson tags: CBSE Class 10 AI Project Cycle Notes - 6, Neural Networks
Back to: Class 10 Artificial Intelligence Notes
Spread the love