Andrew Ng (Introduction to Deep Learning) C1Week1

• 課程介紹
• Neural Networks and Deep Learning(
• NN其他的基礎，妳可以辨識出一隻貓

• Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
• 調整參數、正則化
• Momentum Amest Prop

• Structuring your Machine Learning project
• Cross-validation sets and test sets
• end to end
• Convolutional Neural Networks(CNN)
• Natural Language Processing: Building sequence models
• Recurrent Neural Network(RNN)
• LSTM(Long short term memory models)

•   Deep Learning Apply to SL
• What's a neural network?
• Housing Price Prediction 
• Price>0(Simple NN)
• Size(Input) ->Neural(一個這個O) ->Price(output)
• Relu function = max(0,y)
• 給 input x and output y middle will figure out themselves 
• 給足各地X,Y 可以訓練出好的model

•  Supervised Learning with Neural Networks  
• Real Estate(NN)
• Online Advertising (NN)
• Photo tagging(CNN)
• Speech recognition(RNN) sequence date
• Machine Transiation(RNN) sequence data
• Autonomous driving

• Why is Deep Leaning taking off

• M =Training set的大小
• if m太小，優劣很難說
• Sigomal function >Relu function (Gradient decent跑比較快) 

• About this course
• Introduction 
• Basics of NN progrmming 
• One hidden layer NN
• Deep NN

RNN are a model type / The iterative process of developing DL system is a completely separate concept.