Machine Learning Exposed!

James WeaverDeveloper Advocate

@JavaFXpert

#s1p #springone

From introductory video in Machine Learning course (Stanford University & Coursera) taught by Andrew Ng.

@JavaFXpert

Self-driving cars

@JavaFXpert

Generating image descriptions

Credit: Andrej Karpathy, Li Fei-Feihttp://cs.stanford.edu/people/karpathy/deepimagesent/

@JavaFXpert

Supervised Learning

@JavaFXpert

Supervised learning regression problem(from Andrew Ng’s Machine Learning course)

@JavaFXpert

Unsupervised Learning

@JavaFXpert

Unsupervised learning finds structure in unlabeled data

(e.g. market segment discovery, and social network analysis)

@JavaFXpert

Reinforcement Learning

@JavaFXpert

AlphaGo is a recent reinforcement learning success story

Source: https://gogameguru.com/i/2016/03/AlphaGo-Lee-Sedol-game-3-game-over.jpg@JavaFXpert

Supervised learning classification problem(using the Iris flower data set)

Sepal length

Sepal width

Petal length

Petal width

Species

5.1 3.5 1.4 0.2 Iris setosa

4.9 3.0 1.4 0.2 Iris setosa

7.0 3.2 4.7 1.4 Iris versicolor

6.4 3.2 4.5 1.5 Iris versicolor

6.3 3.3 6.0 2.5 Iris virginica

5.8 3.3 6.0 2.5 Iris virginica

Features LabelsTraining / test data

@JavaFXpert

Iris data classified in four

dimensions

Sepal length

Sepal width

Petal length

Petal width

Species

5.1 3.5 1.4 0.2 Iris setosa

4.9 3.0 1.4 0.2 Iris setosa

7.0 3.2 4.7 1.4 Iris versicolor

6.4 3.2 4.5 1.5 Iris versicolor

6.3 3.3 6.0 2.5 Iris virginica

5.8 3.3 6.0 2.5 Iris virginica

Credit: Tal Galilihttps://cran.r-project.org/web/packages/dendextend/vignettes/Cluster_Analysis.html

Features LabelsTraining / test data

@JavaFXpert

Modeling the brain works well with machine learning(ya think?)

(inputs)

(output)

@JavaFXpert

Anatomy of an Artificial Neural Network(aka Deep Belief Network when multiple hidden layers)

Input layer Output layerHidden layers

Neurons

Synapses@JavaFXpert

Neural net visualization app (uses Spring and DL4J)https://github.com/JavaFXpert/visual-neural-net-serverhttps://github.com/JavaFXpert/ng2-spring-websocket-client

@JavaFXpert

Entering feature values for prediction (classification)

@JavaFXpert

Visual Neural Network application architectureSpring makes REST services and WebSockets easy as π

Neural Net Model Listener

HTML5 client (angular 2 & visjs)

DeepLearning4j library

Neural net graph (WebSocket)

Prediction REST service

prediction & activations

Model Selection Handler

connect & subscribe(WebSocket)

@JavaFXpert

@JavaFXpert

The app leverages machine learning library found at deeplearning4j.org

Simple neural network trained for XOR logic

forward propagation

@JavaFXpert

Feedforward calculations with XOR exampleFor each layer:

(1 x 8.54) + (0 x 8.55) = 8.54

1 / (1 + e -4.55) = 0.99Use sigmoid activation function:

Multiply inputs by weights:Add bias: 8.54 + (-3.99) = 4.55

@JavaFXpert

Simple neural network trained for XOR logic

back propagation (minimize cost function)

@JavaFXpert

Back propagationUses gradient descent to iteratively minimize the cost function

@JavaFXpert

@JavaFXpert

kaggle.com is a great website for data science and machine learning enthusiasts

@JavaFXpert

Let’s use a dataset from kaggle.com to train a neural net on speed dating

Identify features and label we’ll use in the modelLet’s use 65% of the 8378 rows for training and 35% for testing

@JavaFXpert

Code that configures our speed dating neural net

@JavaFXpert

Trying our new speed dating neural net example

@JavaFXpert

Making predictions with our speed dating neural netNote that input layer neuron values are normalized

@JavaFXpert

Machine Learning Exposed!

James WeaverDeveloper Advocate

@JavaFXpert

#s1p #springone

(Thanks for your kind attention)