Udacity Offering Free TensorFlow 2.0 AI Course

Nextbigfuture interviewed Dr. Mat Leonard of the Udacity School of AI. Udacity and Google have created a free course to help software developers to learn code AI applications.

There will be 3 to 4 lessons released every few weeks. The 14 lessons can be completed over 2 months. The coursework will take about 5 to 10 hours per week.

Go to this link to register for the free course.

You can see the Nanodegree programs and free courses in the Udacity School of AI here:

Udacity works with industry leaders like Google, IBM and Amazon to build job-relevant education in fields like Natural Language Processing and Deep Learning.

The goal of the Udacity courses are to get you building state-of-the-art AI applications as fast as possible, without requiring a background in math. If you can code, you can build AI with TensorFlow. You’ll get hands-on experience using TensorFlow to implement state-of-the-art image classifiers and other deep learning models. You’ll also learn how to deploy your models to various environments including browsers, phones, and the cloud.

In 2016, Udacity released the very first free course on TensorFlow in collaboration with Google. Since then, over 400,000 students have enrolled in the course and joined the AI revolution. We’re excited to release an all-new version of this free course featuring the just-announced alpha release of TensorFlow 2.0: Intro to TensorFlow for Deep Learning. This update makes AI even more accessible to everyone, and we’ve again worked directly with the deep learning experts at Google to ensure you’re learning the very latest skills to utilize TensorFlow.

Google has standardized around Keras as the main API. This course focuses on Python for development.

The course will use the Google CoLab. This is cloud based access to GPUs for AI programming. Jupyter notebooks will also be used in the free course.

The Jupyter Notebook is an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text.

Students will learn about Convolutional Neural Networks (CNN), data augmentation, and transfer learning. These techniques to increase the accuracy of deep neural networks and you’ll get hands-on experience by optimizing and testing your neural networks on different image datasets. Later in the course, you’ll learn how to deploy your trained models on browsers, Android, iOS, and embedded devices like the Raspberry Pi, as well as how to perform object detection, and much more.

This free course is part of Udacity’s School of AI, a set of free courses and Nanodegree programs designed by and for software developers.

SOURCES- Udacity, Interview Dr Matt Leonard

Written By Brian Wang

3 thoughts on “Udacity Offering Free TensorFlow 2.0 AI Course”

  1. You have to suggest a better alternative, not just complain about the one that has been used in that context since the 1990s.

  2. Can we please avoid using pretentious buzzwords like ecosystem? You’re turning this into another Blockchain commercial.

  3. The TF 2.0 release is pretty big for a lot of reasons, so pushing free education by famous DL/ML guys is important. The Coral dev board allowing mere mortals access to a cut down TPU is actually big, but not for what you think.

    TF 2.0 now includes TF Federated and TF Privacy, These two together basically mean safe-ish aquistion of model updates from edge devices, for collection and reintegration into a cloud hosted machine learning model, for redelivery out back to edge devices. This is apparently already being used for android keyboard, where they want the model updates, and not the inputted characters directly (because people don’t like the idea of google seeing every keystroke when not using google properties).

    This effectively means they are expecting a cloud to edge distribution model, which closely aligns with some current work on IoT/Edge computing using a Raspberry Pi and docker containers. While the edge docker container ecosystem is not quite there yet (rancher and k3s are starting to cover this, but actual deployment stuff including IoT device management is not yet integrated), it’s easy to see cloud companies moving to this model, especially if those containers are increasingly FaaS-like. It would simplify hauling neural network model updates to things like networked surveillance cameras for instance.

Comments are closed.