In this course, I accomplished 3 assignments about implementing basic deep learning models and learned about various fundamental topics in deep learning.
Assignment 1 | MLP Implementation : Implementing 2-layer neural net with Softmax classfier without Pytorch
Assignment 2 | CNN Architecture Implementation : Implementing ResNet-50 and training with CIFAR-10 dataset
Assignment 3 | Encoder-Decoder Architecture Implementation : Implementing basic U-Net and U-Net with ResNet-based encoder, and conducting image segmentation with Pascal VOC 2012 dataset
In this course, I learned...
- Linear Regression / Logistic Regression
- Learning Theory : Bias-Variance Tradeoff, Under/Overfitting, Regularization, Model Selection
- Neural Networks and Backpropagation
- Convolutional Neural Networks (CNN) and CNN architectures : LeNet, AlexNet, VGG, GoogLeNet, ResNet...
- Training Neural Networks I : Activation functions, Data Preprocessing, Weight Initialization, Batch Normalization, LR scheduling, Hyperparameter Optimization
- Training Neural Networks II : Optimizer (SGD, Momentum, Nesterov Momentum), Regularization Techniques, Transfer Learning
- Encoder-Decoder Architecture
- Recurrent Neural Networks (RNN)
- Visualizing and Understanding of CNNs : Saliency Maps, Guided Backpropagation, Gradient Ascent, DeepDream, Feature Inversion, CAM
- Generative Adversarial Networks (GAN)
- Attention, Transformers and Vision Transformers