Neural Networks

Google Tool

Part 1:  Conceptual understanding  and practice

https://developers.google.com/machine-learning/crash-course/feature-crosses/video-lecture?hl=en

https://developers.google.com/machine-learning/crash-course/introduction-to-neural-networks/anatomy?hl=en


https://developers.google.com/machine-learning/crash-course/reducing-loss/learning-rate

https://developers.google.com/machine-learning/crash-course/reducing-loss/playground-exercise


https://developers.google.com/machine-learning/crash-course/training-and-test-sets/playground-exercise?hl=en

         batch size The number of examples in a batch. For instance, if the batch size is 100, then the model processes 100 examples per iteration."

         https://developers.google.com/machine-learning/glossary?hl=en#epoch


https://developers.google.com/machine-learning/crash-course/regularization-for-simplicity/video-lecture?hl=en


Task 4: Continue experimenting by adding or removing hidden layers and neurons per layer. Also feel free to change learning rates, regularization, and other learning settings. What is the smallest number of neurons and layers you can use that gives test loss of 0.177 or lower?


Part 2:  Program exercise


https://developers.google.com/machine-learning/crash-course/california-housing-data-description

https://developers.google.com/machine-learning/crash-course/introduction-to-neural-networks/programming-exercise

Some Hight-Light

         --> from tensorflow.keras import layers

         --> model = tf.keras.models.Sequential()

         --> learning_rate = 0.01

             epochs = 15

             batch_size = 1000

         -->

             model.add(tf.keras.layers.Dense(units=20,

                                 activation='relu',

                                 name='Hidden1'))



  Other reference 

https://gist.github.com/SupremeLeaf/b94f2580e96fbef4e74570eecf5e5a90

https://www.youtube.com/watch?v=p7n7DVudm78