Convolutional and Recurrent Neural Networks
An Introduction
Aditya Srivastav
June 22, 2024
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 1 / 16
Introduction
Brief introduction to CNNs: A Convolutional Neural Network (CNN)
is a deep learning algorithm that can take in an input image, assign
importance to various aspects/objects in the image, and differentiate
one from the other. Widely used in image and video recognition.
Figure: CNN
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 2 / 16
Introduction
Why we need CNN.
Figure: Image 16mp
Importance in the field of computer vision.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 3 / 16
Introduction
Convolutional Operation in CNN:
Figure: convolution Operation
Padding in CNN.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 4 / 16
CNN Stride
Stride in CNN
Figure: Stride Examples
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 5 / 16
CNN
Max Pooling in CNN
Figure: Stride Examples
Fully Connected layer in CNN.
Figure: Fully Connected Layer in CNN
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 6 / 16
Architecture of CNNs
Figure: CNN Architecture
convolutional layers.
pooling layers.
Fully connected layers.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 7 / 16
Applications of CNNs
Image classification.
Object detection.
Other applications in various domains.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 8 / 16
RNN
Whar is RNN: A Recurrent Neural Network is a specialized type of
artificial neural network designed to handle sequential or time series
data. Unlike feedforward networks, RNNs maintain an internal state
(memory) that allows them to remember past inputs and use that
information to influence their current output.
Figure: RNN
Why we Need RNN
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 9 / 16
Types of RNN
One to One
One to Many
Many to One
Many to Many( equal I/O)
Many to Many(unequal I/O)
Figure: Types of RNN
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 10 / 16
LSTM and Architecture of RNN
Architecture of RNN
Figure: Architecture of RNN
LSTM (Long Short Term Memory)
Why we Need LSTM :
Vanishing Gradient Descent in RNN
It can remember context through a long time, and it has both
long-term and short-term memory.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 11 / 16
LSTM
LSTM Cell
Gates
Forget Gate
Input Gate
Output Gate
Figure: LSTM Cell
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 12 / 16
Back Propogation
Backpropogation in RNN
Figure: BackPrpogation in RNN
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 13 / 16
Applications of RNN
Natural Language Processing (NLP).
Speech Recognition.
Time Series Prediction.
Music Composition.
Signal Classification.
Predictive Maintenance.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 14 / 16
Conclusion
In this presentation i have covered the basics and the architecture of the
cnn and learned about the use cases of the convolution neural networks in
diffrent feild such as image classification , object detection and other
diffrent domains and then learned about Recurrent neural network and
there diffrent useCases.
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 15 / 16
Conclusion
Thank you Sir!
Aditya Srivastav Convolutional and Recurrent Neural Networks June 22, 2024 16 / 16