It covers Assignments done in Deep Learning Course in Indraprastha Institute of Information Technology, Delhi (IIITD) and Udacity NanoDegree Intro to Deep Learning uisng Pytorch.
| S.NO | TOPICS | PROJECT NAME |
|---|---|---|
| 01. | PTA for AND,OR,NOT and XOR and Madeline Implementation from scratch | Implementing PTA and Madeline |
| 02. | From scratch implementation of Back Propagation with optimizers Momentum, NAG, AdaGrad, RMSProp, Adam and initializations He, Xavier and Regularization using L1, L2 and Dropout. No Deep Learning library used. | Backpropagation Optimizers Regularization from scratch |
| 03. | CNN Implementation | Convolutional Neural Networks |
| 04. | Implemented Papers Show, Attend and Tell: Neural Image Caption Generation with Visual Attention and Interactive Attention Networks for Aspect-Level Sentiment Classification | Attention models |
| S.NO | TOPICS | PROJECT NAME |
|---|---|---|
| 01. | Implementing Gradient Descent over a set of random data | 1_GradientDescent |
| 02. | Simple Neural Network and common functions like tensor.view() tensor.reshape() tensor.shape tensor.rand | 2_Simple Neural Network and Random Functions |
| 03. | Creating Multi-Layer Neural Network and converting numpy array to tensors | 3_Multi Layer Neural Networks & numpy to torch |
| 04. | Digit Classification dataset using softmax and matrix multiplication(NO TRAINING) | 4_Digit Classification with Softmax (NO TRAINING) |
| 05. | pytorch nn module for complex neural networks , using torch.nn.functional | 5_Building networks with Pytorch - nn Module |
| 06. | other Activations , Neural Network using Relu and nn.Sequential , Changing weights and biases , using OrderedDict to name individual layers | 6_Relu Activation neural network and nn.Sequential |
| 07. | Training network over Digit Classification - loss calculation-criterion , Autograd , update weights using Pytorch -optim | 7_Training Neural Network |
| 08. | Training neural network to classify Fashion-MNIST | 8_Classifying Fashion-MNIST |
| 09. | Test over Test data , overfitting, regularization using Dropout and Accuracy Calculation | 9_Fashion MNIST - INFERENCE AND VALIDATION |
| 10. | Saving models using state_dict and training later on | _10_Saving and Loading Models |
| 11. | Making filters and visualising CNN | convulution-Neural-Network |
| 12. | Transfer learning | 12_CATS_VS_DOG_CLASSIFICATION_TRANSFER_LEARNING |
| 13. | STYLE TRANSFER | Style_Transfer |