Machine Learning with Python : for PC, Raspberry Pi, and MaixDuino

Most people are increasingly confronted with the applications of Artificial Intelligence (AI). Music or video ratings, navigation systems, shopping advice, etc. are based on methods that can be attributed to this field.The term Artificial Intelligence was coined in 1956 at an international conference known as the Dartmouth Summer Research Project. One basic approach was to model the functioning of the human brain and to construct advanced computer systems based on this. Soon it should be clear how the human mind works. Transferring it to a machine was considered only a small step. This notion proved to be a bit too optimistic. Nevertheless, the progress of modern AI, or rather its subspecialty called Machine Learning (ML), can no longer be denied.In this book, several different systems will be used to get to know the methods of machine learning in more detail. In addition to the PC, both the Raspberry Pi and the Maixduino will demonstrate their capabilities in the individual projects. In addition to applications such as object and facial recognition, practical systems such as bottle detectors, person counters, or a “talking eye” will also be created.The latter is capable of acoustically describing objects or faces that are detected automatically. For example, if a vehicle is in the field of view of the connected camera, the information "I see a car!" is output via electronically generated speech. Such devices are highly interesting examples of how, for example, blind or severely visually impaired people can also benefit from AI systems. Search... Machine Learning with Python All rights reserved. Contents Cautionary Notices Program Downloads 1 • Introduction 1.1 "Super Intelligence" in three steps? 1.2 How machines can learn 2 • A Brief History of ML and AI 3 • Learning from "Big Data" 4 • The Hardware Base 5 • The PC as Universal AI Machine 5.1 The computer as a programming center 6 • The Raspberry Pi 6.1 The Remote Desktop 6.2 Using smartphones and tablets as displays 6.3 FileZilla 6.4 Pimp my Pi 7 • Sipeed Maix, aka "MaixDuino" 7.1 Small but mighty: the performance figures of the MaixDuino 7.2 A wealth of applications 7.3 Initial start-up and functional test 7.4 Power supply and stand-alone operation 8 • Programming and Development Environments 8.1 Thonny — a Python IDE for beginners and intermediates 8.2 Thonny as a universal IDE for RPi and MaixDuino 8.3 Working with files 8.4 Thonny on the Raspberry Pi 8.5 Tips for troubleshooting the Thonny IDE 8.6 The MaixPy IDE 8.7 A MicroPython interpreter for MaixDuino 8.8 The Flash tool in action 8.9 Machine Learning and interactive Python 8.10 Anaconda 8.11 Jupyter 8.12 Installation and Start-Up 8.13 Using MicroPython Kernels in Jupyter 8.14 Communication setup to the MaixDuino 8.15 Kernels 8.16 Working with Notebooks 8.17 All libraries available? 8.18 Using Spyder for Python Programming 8.19 Who's programming who? 9 • Python in a Nutshell 9.1 Comments make your life easier 9.2 The print() statement 9.3 Output to the display 9.4 Indentations and Blocks 9.5 Time Control and Sleep 9.6 Hardware under control: digital inputs and outputs 9.7 For vital values: variables and constants 9.8 Numbers and variable types 9.9 Converting number types 9.10 Arrays as a basis for neural networks 9.11 Operators 9.12 Conditions, branches and loops 9.13 Trial and error — try and except 10 • Useful Assistants: Libraries! 10.1 MatPlotLib as a graphics artist 10.2 The math genius: Numpy 10.3 Data-mining using Pandas 10.4 Learning and visualization using Scikit, Scipy, SkImage & Co. 10.5 Machine Vision using OpenCV 10.6 Brainiacs: KERAS and TensorFlow 10.7 Knowledge transfer: sharing the learning achievements 10.8 Graphical representation of network structures 10.9 Solution of the XOR problem using KERAS 10.10 Virtual environments 11 • Practical Machine Learning Applications 11.1 Transfer functions and multilayer networks 11.2 Flowers and data 11.3 Graphical representations of data sets 11.4 A net for iris flowers 11.5 Training and testing 11.6 What's blossoming here? 11.7 Test and learning behavior 12 • Recognition of Handwritten Numbers 12.1 "Hello ML" — the MNIST data set 12.2 A neural network reads digits 12.3 Training, tests and predictions 12.4 Live recognition of digits 12.5 KERAS can do even better! 12.6 Convolutional networks 12.7 Power training 12.8 Quality control — an absolute must! 12.9 Recognizing live images 12.10 Batch sizes and epochs 12.11 MaixDuino also reads digits 13 • How Machines Learn to See: Object Recognition 13.1 TensorFlow for Raspberry Pi 13.2 Virtual environments in action 13.3 Using a Universal TFlite Model 13.4 Ideal for sloths: clothes-sorting 13.5 Construction and training of the model 13.6 MaixDuino recognizes 20 objects 13.7 Recognizing, counting and sorting objects 14 • Machines Learn to Listen and Speak 14.1 Talk to me! 14.2 RPi Learns to talk 14.3 Talking instruments 14.4 Sorry, didn't get you ... 14.5 RPi as a ChatBot 14.6 From ELIZA to ChatterBots 14.7 The Talking Eye 14.8 An AI Bat 15 • Facial Recognition and Identification 15.1 The right to your own image 15.2 Machines recognize people and faces 15.3 MaixDuino as a Door Viewer 15.4 How many guest were at the party? 15.5 Person-detection alarm 15.6 Social minefields? — face identification 15.7 Big Brother RPi: face identification in practice 15.8 Smile, please ;-) 15.9 Photo Training 15.10 "Know thyself!" ... and others 15.11 A Biometric scanner as a door opener 15.12 Recognizing gender and age 16 • Train Your Own Models 16.1 Creation of a model for the MaixDuino 16.2 Electronic parts recognition with the MaixDuino 16.3 Performance of the trained network 16.4 Field test 16.5 Outlook: Multi-object detectors 17 • Dreams of the Future: from KPU to Neuromorphic Chips 18 • Electronic Components 18.1 Breadboards 18.2 Wires and jumpers 18.3 Resistors 18.4 Light-emitting diodes (LEDs) 18.5 Transistors 18.6 Sensors 18.7 Ultrasound range finder 19 • Troubleshooting 20 • Buyers Guide 21 • References; Bibliography Index