Scribd
Upload
7 views3 pages

Complete ANN DeepLearning Notes

Artificial Neural Networks (ANN) mimic the human brain's structure and function, consisting of interconnected neurons organized into layers. They have advantages like parallel processing and fault tolerance but face challenges such as their black-box nature and high computational requirements. Deep Learning, a subset of Machine Learning, utilizes deep neural networks for tasks like image and speech recognition, offering high accuracy but requiring large datasets and expensive training.

Uploaded by

mohitgupta23604
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
7 views3 pages

Complete ANN DeepLearning Notes

Artificial Neural Networks (ANN) mimic the human brain's structure and function, consisting of interconnected neurons organized into layers. They have advantages like parallel processing and fault tolerance but face challenges such as their black-box nature and high computational requirements. Deep Learning, a subset of Machine Learning, utilizes deep neural networks for tasks like image and speech recognition, offering high accuracy but requiring large datasets and expensive training.

Uploaded by

mohitgupta23604
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

ARTIFICIAL NEURAL NETWORK (ANN)

Artificial Neural Network is inspired by the biological neural network of the human brain.

It consists of interconnected neurons (nodes) organized into layers: input layer, hidden layer, and
output layer.

Biological vs Artificial:

Dendrites → Inputs

Cell nucleus → Node

Synapse → Weights

Axon → Output

ANN Working:

Input is multiplied by weights, summed with bias, and passed through an activation function.

Advantages:

- Parallel processing

- Fault tolerance

- Handles incomplete data

Disadvantages:

- Black-box nature

- Requires high computation

- No fixed structure

PERCEPTRON

Perceptron is a single-layer neural network used for binary classification.

It forms a linear decision boundary.

Working:

Weighted sum of inputs → Activation function → Output

Limitations:

- Cannot solve XOR problem (non-linear)


MULTILAYER PERCEPTRON (MLP)

MLP contains multiple hidden layers and can solve non-linear problems.

It uses feedforward and backpropagation.

Features:

- Handles complex patterns

- Uses activation functions like ReLU, Sigmoid

BACKPROPAGATION

Backpropagation is used to train neural networks by updating weights.

Steps:

1. Forward propagation

2. Error calculation (MSE)

3. Backward propagation

4. Weight update using gradient descent

The process continues until error is minimized.

DEEP LEARNING

Deep Learning is a subset of Machine Learning using deep neural networks.

Features:

- Automatic feature extraction

- Works with large data

- High accuracy

Types:

- Feedforward Neural Network (FNN)

- Convolutional Neural Network (CNN)

- Recurrent Neural Network (RNN)

- Deep Neural Network (DNN)


- Deep Belief Network (DBN)

Machine Learning vs Deep Learning:

ML → Small data, less computation

DL → Large data, high computation

Advantages:

- Best performance

- Reduces manual work

Disadvantages:

- Needs large data

- Expensive training

Applications:

- Image Recognition

- Speech Recognition

- NLP

- Healthcare

- Finance

- Self-driving cars

You might also like