Python and Deep Learning: Special Topics

Rajeshwari Sai Aishwarya Puppala

Student ID: 16298162

Class ID: 35

Deep Learning-In class programming:4

Objectives

Follow the instruction below and then report how the performance changed. (apply all at once)

Did the performance change?

2.predict the first 4 image of the test data. Then, print the actual label for those 4 images (label means the probability associated with them) to check if the model predicted correctly or not

3.Visualize Loss and Accuracy using the history object

Import Data

• Import the necessary Packages required

• Import the cifar dataset and load all of the train and test data

Normalize and Encoding

Model1

• Create a convolution layer with activation "Relu" input shape (3,32,32)
• Add a dropout layer with 0.2
• Add a Conv2d Hidden layer with activation function "Relu" and with 32 neurons
• Do the max pooling with (2,2) size.
• Now Flatten the size of the input
• Add a dense hidden layer with activation function "Relu"
• The activation function in the output layer is "softmax" because the target is multi-classification
• The hyperparameters are as follows epochs=2, Lrrate=0.01, optimizer is "SGD"

Accuracy: 70.3%

Model 2

• Convolutional input layer, 32 feature maps with a size of 3×3 and a rectifier activation function.
• Dropout layer at 20%.Convolutional layer, 32 feature maps with a size of 3×3 and a rectifier activation function.
• Max Pool layer with size 2×2.Convolutional layer, 64 feature maps with a size of 3×3 and a rectifier activation function.
• Dropout layer at 20%.
• Convolutional layer, 64 feature maps with a size of 3×3 and a rectifier activation function.
• Max Pool layer with size 2×2.Convolutional layer, 128feature maps with a size of 3×3 and a rectifier activation function.
• Dropout layer at 20%.
• Convolutional layer,128 feature maps with a size of 3×3 and a rectifier activation function.
• Max Pool layer with size 2×2.Flatten layer.Dropout layer at 20%.
• Fully connected layer with 1024units and a rectifier activation function.
• Dropout layerat 20%.
• Fully connected layer with 512units and a rectifier activation function.
• Dropoutlayer at 20%.Fully connected output layer with 10 units and a Softmax activation function

HyperParameters code

Accuracy: 39.8%

Accuracy

Loss

Predict 4 images from the Test dataset

• Code for predicting the results of 4 images in the test data set with the model already got trained
• As you can see the result the second vector is one hot encoded
• It shows 1 that is the class it is representing
• Check the same index value in the predicted values it should be highest compared to other 9 values then it is predicted correctly.
• The third image has 1 for the 9th class and see the values in the predicted values, it is the highest values.
• The 3rd image got predicted correctly

Conclusion

• The Model1 has better accuracy compared to model 2.