DL Lab 2 - SASLEENREZA/Python_DeepLearning GitHub Wiki

Team ID: 15

Name : Sasleen Reza Shaik, Class ID : 27

Name : Navya Ramya Sirisha, Class ID : 24

Video link : https://drive.google.com/file/d/1POcel8e8Yu92c1tZn4Emm5MPmet4DgKq/view?usp=sharing

Introduction :

The main aim of the lab assignment is to create an exposure to Deep Learning Concepts and apply CNN, LSTM and RNN models for Text and Image Classification.

Objective :

The assignment is divided into four tasks which focuses to make one familiar with deep learning concepts listed above:

• Implement the text classification with CNN model.
• Implement the text classification with RNN/LSTM model.
• Comparing the results of CNN and RNN/LSTM models, for the text classification for same dataset and to describe which model is best for the text classification.
• Implement the image classification with CNN model.

Program-1 : Text classification with CNN

For this task we took Aviation Dataset from Kaggle and then we loaded the data into the model. After loading the data we divided the data into training and testing data and then we evaluated the data using CNN model by applying maxpooling, dense and convolutional layers. And then we fitted the data into the model and plotted the graphs for loss and accuracy obtained.

Program-2 : Text classification with RNN/LSTM

For this task we took Aviation Dataset from Kaggle and then we loaded the data into the model. After loading the data we divided the data into training and testing data and then we evaluated the data using RNN/LSTM model by applying maxpooling, dense and convolutional layers. And then we fitted the data into the model and plotted the graphs for loss and accuracy obtained.

Program-3 : Comparison

Model CNN RNN LSTM

Accuracy 89.65% 84.35% 93.27%

Loss 6.69% 37.25% 18.08%

LSTM>CNN>RNN

Program-4 : Image classification with CNN

For this model we took CIFAR10 dataset and then we divided the data into training and testing data and then we evaluated the data by applying maxpooling, dense and convolutional layers. And then we fitted the data into the model and plotted the graphs for loss and accuracy obtained.

Conclusion:

From our observation we figured out that LSTM model works best for text analysis than any other models.