Preface
- This book uses Python as the programming language, uses as few external libraries as possible, and implements deep learning programs from scratch.
This book starts with a simple machine learning problem and finally realizes a system that can recognize images with high accuracy. - This book focuses on image recognition and mainly learns the techniques required for image recognition using deep learning. Natural language processing or speech recognition is not the subject of discussion in this book.
- This book does not introduce the use of Caffe, TensorFlow, Chainer, Theano and other deep learning frameworks.
- This book does not introduce detailed theories of deep learning, especially related detailed theories of neural networks, and does not introduce parameter tuning related content for improving recognition accuracy.
- This book does not involve the latest research and theoretical details, but after reading this book, readers should be able to further read the latest papers or technical books related to neural networks.
1. What is Python
-
Python is a simple, easy-to-read, easy-to-remember programming language, and it is open source. For those who are new to programming, Python is the most suitable language.
-
Companies at the forefront of the IT industry such as Google, Microsoft, and Facebook often use Python.
-
In the field of science, especially in the field of machine learning and data science, Python is widely used. Python has an unshakable position in the field of data science by virtue of excellent numerical calculation and statistical analysis libraries such as NumPy and SciPy. There are also many scenarios where Python is used in the deep learning framework, such as Caffe, TensorFlow, Chainer, Theano and other famous deep learning frameworks that provide Python interfaces. Therefore, learning Python is of great benefit to using deep learning frameworks.
2. Python installation
2.1 Python version
I am using Python 3.7.6.
- Viewing method:
Shortcut key: Win+R
click'OK' to open the terminal. - Enter the
python --versioncommand in the terminal. The
output shows the version information of the installed Python.
2.2 External libraries used
This book mainly uses NumPy and Matplotlib .
- NumPy is a library for numerical calculations , providing many advanced mathematical algorithms and convenient matrix manipulation methods. This book will use these convenient methods to effectively promote the realization of deep learning.
- Matplotlib is a library for drawing graphs . Using Matplotlib can visualize the experimental results and visually confirm the data during the deep learning run.
2.3 Anaconda release version
I use Anaconda