Machine Learning with Swift

Book Image

Machine Learning with Swift

By : Jojo Moolayil, Alexander Sosnovshchenko, Oleksandr Baiev

Book Image

Machine Learning with Swift

By: Jojo Moolayil, Alexander Sosnovshchenko, Oleksandr Baiev

Overview of this book

Machine learning as a field promises to bring increased intelligence to the software by helping us learn and analyse information efficiently and discover certain patterns that humans cannot. This book will be your guide as you embark on an exciting journey in machine learning using the popular Swift language. We’ll start with machine learning basics in the first part of the book to develop a lasting intuition about fundamental machine learning concepts. We explore various supervised and unsupervised statistical learning techniques and how to implement them in Swift, while the third section walks you through deep learning techniques with the help of typical real-world cases. In the last section, we will dive into some hard core topics such as model compression, GPU acceleration and provide some recommendations to avoid common mistakes during machine learning application development. By the end of the book, you'll be able to develop intelligent applications written in Swift that can learn for themselves.

Title Page

Packt Upsell

Contributors

Preface

Free Chapter

Getting Started with Machine Learning

Getting Started with Machine Learning

The motivation behind ML

Applications of ML

Using ML to build smarter iOS applications

Getting to know your data

Choosing a model

Classification – Decision Tree Learning

Classification – Decision Tree Learning

Machine learning toolbox

Prototyping the first machine learning app

IPython notebook crash course

Time to practice

Machine learning for extra-terrestrial life explorers

Loading the dataset

Exploratory data analysis

Data preprocessing

Decision trees everywhere

Training the decision tree classifier

How decision tree learning works

Implementing first machine learning app in Swift

Introducing Core ML

K-Nearest Neighbors Classifier

K-Nearest Neighbors Classifier

Calculating the distance

Using instance-based models for classification and clustering

People motion recognition using inertial sensors

Understanding the KNN algorithm

Recognizing human motion using KNN

Reasoning in high-dimensional spaces

Improving our solution

K-Means Clustering

K-Means Clustering

Unsupervised learning

K-means clustering

Implementing k-means in Swift

Clustering objects on a map

Choosing the number of clusters

K-means clustering – problems

Image segmentation using k-means

Association Rule Learning

Association Rule Learning

Seeing association rules

Defining data structures

Using association measures to assess rules

Decomposing the problem

Generating all possible rules

Finding frequent item sets

The Apriori algorithm

Implementing Apriori in Swift

Running Apriori

Running Apriori on real-world data

The pros and cons of Apriori

Building an adaptable user experience

Linear Regression and Gradient Descent

Linear Regression and Gradient Descent

Understanding the regression task

Introducing simple linear regression

Feature scaling

Feature standardization

Implementing multiple linear regression in Swift

Fixing linear regression problems with regularization

Linear Classifier and Logistic Regression

Linear Classifier and Logistic Regression

Revisiting the classification task

Implementing logistic regression in Swift

Predicting user intents

Choosing the regression model for your problem

Bias-variance trade-off

Neural Networks

Neural Networks

What are artificial NNs anyway?

Building the neuron

Building the network

Building a neural layer in Swift

Using neurons to build logical functions

Implementing layers in Swift

Training the network

Basic neural network subroutines (BNNS)

Convolutional Neural Networks

Convolutional Neural Networks

Understanding users emotions

Introducing computer vision problems

Introducing convolutional neural networks

Pooling operation

Convolution operation

Building the network

Training the network

Training the CNN for facial expression recognition

Environment setup

Deep learning frameworks

Loading the data

Splitting the data

Data augmentation

Creating the network

Plotting the network structure

Training the network

Making predictions

Saving the model in HDF5 format

Converting to Core ML format

Visualizing convolution filters

Deploying CNN to iOS

Natural Language Processing

Natural Language Processing

NLP in the mobile development world

Word Association game

Python NLP libraries

Textual corpuses

Common NLP approaches and subtasks

Distributional semantics hypothesis

Word vector representations

Autoencoder neural networks

Word2Vec in Gensim

Vector space properties

iOS application

Word2Vec friends and relatives

Where to go from here?

Machine Learning Libraries

Machine Learning Libraries

Machine learning and AI APIs

General-purpose machine learning libraries

Inference-only libraries

Speech recognition

Computer vision

Low-level subroutine libraries

Choosing a deep learning framework

Optimizing Neural Networks for Mobile Devices

Optimizing Neural Networks for Mobile Devices

Delivering perfect user experience

Calculating the size of a convolutional neural network

Lossless compression

Compact CNN architectures

Preventing a neural network from growing big

Lossy compression

An example of the network compression

Best Practices

Mobile machine learning project life cycle

Machine learning gremlins

Recommended learning resources

Index

Customer Reviews

5 star

0

4 star

0

3 star

0

2 star

0

1 star

0

The pros and cons of Apriori

The pros of Apriori are as follows:

This is the most simple and easy-to-understand algorithm among association rule learning algorithms
The resulting rules are intuitive and easy to communicate to an end user
It doesn't require labeled data as it is fully unsupervised; as a result, you can use it in many different situations because unlabeled data is often more accessible
Many extensions were proposed for different use cases based on this implementation—for example, there are association learning algorithms that take into account the ordering of items, their number, and associated timestamps
The algorithm is exhaustive, so it finds all the rules with the specified support and confidence

The cons of Apriori are as follows:

If the dataset is small, the algorithm can find many false associations that happened simply by chance. You can address this issue by evaluating obtained rules on the held-out test data for the support, confidence, lift, and conviction values.
As Agrawal...