Deep learning projects from beginner (ex. shallow NN, backprop) to advanced (ex. YOLO, generate jazz music). Topics including deep neural network, improving neural network, building neural network project, convolutional neural network, and sequence model. Started with only using NumPy for shallow NN, then moved on to tensorflow with/without keras for deeper networks.
The primary areas tackled were fundamentals of deep learning, CNN, and RNN. I have also worked on Generative Adversarial Network (GAN) and Reinforcement Learning (RL).
|
---|
_________________________________________________________________________ |
|
_________________________________________________________________________ |
|
|
---|
_________________________________________________________________________ |
|
_________________________________________________________________________ |
|
_________________________________________________________________________ |
|
Neural Networks and Deep Learning
- Introduction to deep learning
- Neural Networks Basics
- Shallow Neural networks
- Deep Neural Networks
Improving Deep Neural Networks
- Practical aspects of Deep Learning
- Optimization algorithms
- Hyperparameter tuning, Batch Normalization and Programming Frameworks
Structuring Machine Learning Projects
- Diagnose and Treat Errors in Machine Learning Model
- Understand Complex Machine Learning Settings and Limitations
- Apply End to End, Tranfer, and Multitask Learning
Convolutional Neural Network
- Foundations of Convolutional Neural Networks
- Deep convolutional models: case studies
- Object detection
- Special applications: Face recognition & Neural style transfer
Sequence Models
- Recurrent Neural Networks with LSTM
- Natural Language Processing & Word Embeddings
- Sequence models & Attention mechanism
Reading these papers allowed me to see how the state of the art models evolve and to gain more indepth understanding of the theory behind deep learning.
- LeNet
- AlexNet
- VGG
- FaceNet (triple loss)
- GoogLeNet (inception)
- Neural Style Transfer
- ResNet
- YOLO (real time object detection)
- YOLO 9000 (lots of crazy impressive numbers)
Visualization of the evolution of CNN architectures
As declared in the description, this collection of projects is mentored by Stanford Professor Andrew Ng through the Coursera Deep Learning Specialization. Thanks to his guidance, I gained balanced knowledge of both the theory and the practical application of deep learning.