This is the code repository for Learning Python Artificial Intelligence by Example [Video], published by Packt. It contains all the supporting project files necessary to work through the video course from start to finish.
Python has surfaced as a dominant language in AI/ML programming because of its simplicity and flexibility. It has great support for open-source libraries such as Scikit-learn and TensorFlow.
Built for rookie AI enthusiasts across four realistic projects, this course covers modern techniques that make up the world of Artificial Intelligence. Dive into your first natural language processing project, build a facial recognition system, and build your very own self driving steering code. You will explore the use of neural networks and deep learning, and how you can train and test sets for feature extraction. You'll be introduced to the Keras deep learning library, which you will use to predict taxi journey times, and to the use of natural language processing to find the most relevant articles in Wikipedia.
By the end of this video course, you will be confident enough to build your own AI projects with Python, and ready to take on more advanced content as you move on.
- Create your own NLP chatbot using Python AI
- Use open source, SaaS and custom-built algorithms to identify faces in pictures and video
- Understand data-mining methods, and how to work with multiple data sets when building a model
- How feature engineering works to get the most value from the data
- Apply open data and deep learning to predict taxi journey times in New York City
- Use convolutional neural networks to determine an appropriate steering angle for a self-driving car
To fully benefit from the coverage included in this course, you will need:
To fully benefit from the coverage included in this course, you will need:
● Prior working knowledge of the Python (version 3) language, including installing libraries using pip and/or conda
● Familiarity with Git and GitHub for source control
● Familiarity with a Linux or Windows operating systems from a command line perspective, including the ability to install OS level libraries (through apt, yum, brew or equivalent)
● (Preferable) Familiarity with Anaconda
● (Optional) Experience with Docker, cloud computing (Google Cloud, AWS or Azure) or Virtual Environments (e.g. VirtualBox or VMWare)
This course has the following software requirements:
This course has the following software requirements:
● An editor like Atom, Sublime Text or PyCharm
● The AWS CLI
It is recommended to use Anaconda, however if you so wish you can manual create environments using Python 3.6
● OpenCV 3.4 and opencv-python
● Tensorflow 1.11 (and pre-requisites if using a GPU)
● Keras 2.2
● geopandas
● Numpy, Pandas and SKLearn (note that these are included in Anaconda)
● Jupyter notebook (note that this is included in Anaconda)
This course has been tested on the following system configuration:
● OS: Mac OSX 10.14.1 (Mojave)
● Processor: Quad Core Intel i5
● Memory: 8GB
● Hard Disk Space: 50GB
● OS: Windows 10, and Ubuntu 16.04 LTS
● Processor: Quad Core Intel i7
● Memory: 16GB
● Hard Disk Space: 50GB
● OS Ubuntu 16.04 LTS
● Processor: 2 core
● Memory: 13GB and 48GB
● Hard Disk Space: 50GB
● Video Card: Nvidia K80
NOTE: The course relies heavily on Jupyter Notebooks so it is essential that you have access to a browser to view and run these.
• Section 2, video 3 relies on the use of AWS Rekognition. To fully complete this video you will need an AWS account with a valid payment card. Note that during development of this course, the AWS charges did not exceed $10, but it is essential that you set up billing alerts and notifications to manage your spend accordingly.
You do not need to complete video 2.3 to get value from the rest of the course
• Due to the nature of deep learning, which is explored in sections 3 and 4, it is advisable to have access to an environment with a suitable Nvidia GPU supported by Tensorflow. This is likely to significantly decrease model training times for these sections.
However, pre-trained models are provided as part of this course, and with enough time (likely to be a few days depending on hardware specification), these can also be trained without a GPU.