AI enthusiasts used to need to invest thousands of dollars on a good GPU in order to get “hands-on” with machine learning, but now a simple web browser is enough. Silicon Valley based nonprofit PracticalAI recently released “practicalAI 2.0,” a platform that includes illustrative machine learning lessons in TensorFlow 2.0 + Keras and has garnered over 23k stars on GitHub.
The platform aims to help users build robust models with custom components and create their own products by training simple and customisable loops. There are tutorials on Jupyter Notebook and Google Colab, Python basic programming, deep learning basics including PyTorch framework and algorithms such as CNN and RNN, advanced level algorithms, and other AI research topics. PracticalAI can be easily run on the Google Colab service with free cloud GPU or TPU provided.
The project author is Goku Mohandas, who previously worked at Apple as an AI researcher. Mohandas says his goal in developing practicalAI is to go beyond classroom study to enable people to learn a more goal-oriented and product-oriented machine learning logic.
The table below summarizes the project’s lesson outline, sectioned into Basic Machine Learning, Production Machine Learning, Advanced Machine Learning and Other AI Topics. The blue text represents lessons that have been put online and users can directly access via Colab; while the black text is lessons that will be released in the future.
To delve deeply into machine learning it’s necessary to gain understanding of the theoretical foundations. Running algorithms on a browser however can allow beginners to experience the charm of AI faster, get started with specific projects, and motivate them to learn more. Mohandas says he hopes the practicalAI platform can contribute to this discovery and education process. Tools to support the platform’s new features are launching this month (January 2020).
For more information please check out the practicalAI Project Page.
Author: Herin Zhao | Editor: Michael Sarazen