Coursera-Deep Learning by Andrew Ng

• Course: https://www.coursera.org/specializations/deep-learning?
• Offered by: https://www.deeplearning.ai/
• Instructor: https://www.andrewng.org/
• My certificates: https://www.coursera.org/account/accomplishments/specialization/55B8USWRBGBQ

Course 01: Neural Networks and Deep Learning

Overview/Topics:

• Neural network, binary classification, logistic regression, Gradient descent, Vectorization
• Python basics wiht numpy, broadcasting
• Understanding Neural network representation, Activation functions and their derivatives, backpropagation, random initialization
• Deep L-layer network, forward propagation, parameters and hyperparameters
• Buildin a deep neural network step by step

Programming Assignments:

• [Week 01: N/A]

• Week 02: Python Basics with numpy optional

• Week 02: Logistic Regression with a Neural Network mindset

• Week 03: Planar data classification with a hidden layer

• Week 04: Building your Deep Neural Network: Step by Step

• Week 04: Deep Neural Network - Application

Research papers that are discussed:

• N/A

Course 02: Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization

Overview/Topics:

• Train/test/dev set and bias, variance
• Regularization: overfitting, underfitting, droupout
• Gradient checking: vanishing/exploding gradients
• Understanding mini-batch gradient descent, bias correction, RMSprop
• Adam optimization, Learning rate decay,
• Hyperperameter tunning process, Batch normalization, Softmax regression
• Understanding deep learning framworks: Tensorflow

Programming Assignments:

• Week 01: Initialization

• Week 01: Regularization

• Week 01: Gradient Checking

• Week 02: Optimization

• Week 03: Tensorflow

Research papers that are discussed:

• N/A

Course 03: Structuring Machine Learning Projects

Overview/Topics:

• Machine Learning strategy: Train/test/dev set distribution
• Comparing human level performance: Avoidable bias
• Error Analysis: Mismatched training dev/test
• Transfer learning, Multitask learning, End-to-end deep learning
• Machine Learning flight simulator

Programming Assignments:

• [Week 01: N/A]

• [Week 02: N/A]

Research papers that are discussed:

• N/A

Course 04: Convolutional Neural Networks

Overview/Topics:

• The basics of computer vision, edge detection, padding, stride, filter, pooling layer
• Build a simple one layer convolutional network
• ResNets, Inception network, Transfer learning, Data augmentation
• Building a Residual networks
• Object detection, IOU, bounding box prediction, non-max suppression, YOLO algorithm, Detection algorithm
• Face recognition and verification: siamese network, triplet loss
• Neural style transfer

Programming Assignments:

• Week 01: Convolutional Model: step by step

• Week 01: Convolutional Model: application

• Week 02: Keras Tutorial (not graded)

• Week 02: Residual Networks

• Week 03: Car detection with YOLO algorithm

• Week 04: Art generation with Neural Style Transfer

• Week 04: Face Recognition

Research papers that are discussed:

• Classic networks: Gradient-based learning applied to document recognition
• Classic networks: ImageNet Classification with Deep Convolutional Neural Networks
• Classic networks: Very Deep Convolutional Networks for Large-Scale Image Recognition
• ResNet: Deep Residual Learning for Image Recognition
• 1x1 convolution: Network In Network
• Inception network: Going Deeper With Convolutions
• Convolutional inplementation with sliding windows: OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks
• Bounding box predictions: You Only Look Once: Unified, Real-Time Object Detection
• Region proposal: R-CNN: Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

Course 05: Sequence Models

Overview/Topics:

• Build a Recurrent Neural Networks(RNN) using Gated Recurrent Unit(GRU) and Long Short Term Memory(LSTM)
• Word Embeddings: word2vec and GloVe word vectors
• Solve NLP problems: Text analysis / Sentiment Analysis
• Sequence to sequence architecture: Beam search, Bleu score, Attention model
• Speech recognition

Programming Assignments:

• Week 01: Building a recurrent neural network - step by step

• Week 01: Dinosaur Island - Character-Level Language Modeling

• Week 01: Jazz improvisation with LSTM

• Week 02: Operations on word vectors - Debiasing

• Week 02: Emojify

• Week 03: Neural Machine Translation with Attention

• Week 04: Trigger word detection

Research papers that are discussed:

• Siamese network: DeepFace: Closing the Gap to Human-Level Performance in Face Verification

• Triplet Loss: FaceNet: A Unified Embedding for Face Recognition and Clustering

• Style transfer: A Neural Algorithm of Artistic Style

• GRU: On the Properties of Neural Machine Translation: Encoder-Decoder Approaches

• GRU: Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

• LSTM: Long Short-Term Memory

• Word embedding: Visualizing Data using t-SNE

• Word embedding: Linguistic Regularities in Continuous Space Word Representations

• Language model: A Neural Probabilistic Language Model

• Negative sampling: Distributed Representations of Words and Phrases and their Compositionality

• Debiasing word embedding: Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings

• RNN: Deep Captioning with Multimodal Recurrent Neural Networks (m-RNN)

• RNN: Show and Tell: A Neural Image Caption Generator

• RNN: Deep Visual-Semantic Alignments for Generating Image Descriptions

• Bleu: BLEU: a Method for Automatic Evaluation of Machine Translation

• NMT: Neural Machine Translation by Jointly Learning to Align and Translate

• Attention model: Show, Attend and Tell: Neural Image Caption Generation with Visual Attention

• CTC: Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

Important Links

• Reading List
• Summaries of papers on deep learning.
• Test your AI/Deep learning skill