Deep Learning Projects

S.No	Projects
01.	Cats vs Dogs Classifier
02.	Rock Paper and Scissors Classifier
03.	American Sign Language Classifier
04.	Humans vs Horses Classifier
05.	GAN
06.	Deep Convolution GAN
07.	Conditional GAN
08.	Wasserstein GAN
09.	Text Classification

P. Cats vs Dogs Classifier

Concepts used:

Transfer Learning (Binary Classification)

1. Transfer Learning (pre-trained InceptionV3 network)
2. Data Augmentation
3. Regularization using Dropouts ( to make network more efficient and prevent overfitting)

↥ back to top

P. Rock Paper and Scissors Classifier

Concepts used:

Multiclass Classification

1. Data Augmentation
2. Regularization using Dropouts ( to make network more efficient and prevent overfitting)

↥ back to top

P. American Sign Language Classifier

Concepts used:

Multiclass Classificaton

1. Data Augmentation
2. Convolution Neural Networks

↥ back to top

P. Humans vs Horses Classifier

Concepts used:

Transfer Learning

1. Transfer Learning (pre-trained InceptionV3 network)
2. Data Augmentation
3. Regularization using Dropouts ( to make network more efficient and prevent overfitting)
4. Callbacks (stops training when reached at certain level)

↥ back to top

P. GAN

Concepts used:

1. Generator
2. Discriminator (Classifier)
3. Linear Transformation
4. Batch Normalization

↥ back to top

P. Deep Convolution GAN

Concepts used:

1. Generator
2. Discriminator (Classifier)
3. Batch Normalization
4. Transpose Convolution

↥ back to top

P. Conditional GAN

Concepts used:

1. Generator
2. Descrimantor
3. BatchNorm
4. Transpose Convolution

↥ back to top

P. Wasserstein GAN

Concepts used:

1. Generator
2. Critic (Discriminator/Classifier)
3. BatchNorm
4. Transpose Convolution
5. Gradient Penalty (To prevent mode collapse)
6. Wasserstein loss

↥ back to top

P. Text Classification

Concepts used:

1. Word Embedding
2. Tokens
3. Pad Sequence
4. Visualization using Tensorflow Projector

↥ back to top