Learning in TensorFlow Seminar
Why should I take this seminar?
This seminar allows you to gain hands-on using TensorFlow and keras to create a machine learning algorithm. This seminar focuses on the steps that you need to perform to create a machine learning algorithm while treating the statistics used to create the algorithms as a black box.
For this and other seminars, the programming language used is Python. We will use TensorFlow, keras, and a couple of other packages.
After completing this seminar, you should be able to:
- Describe the data structures used in machine learning
- Define basic machine learning concepts
- Interpret a program to classify hand-written numbers.
This seminar has three parts. To earn a digital badge, you need to do all three parts: pre-seminar, live seminar, and post-seminar.
You can complete some parts of the seminar, only the live seminar, or only do the pre- or post-seminar, but to earn a digital badge you must complete all three parts.
During the live seminar you should be able to interpret a basic program in Jupyter notebooks that:
- loads the MNIST dataset
- explores the MNIST dataset
- prepares the MNIST dataset
- creates a machine learning model to classify hand-written digits
- reports the accuracy of the model
The pre-seminar module contains:
- A note on machine learning (beta version)
- Links to resources that describe the history of TensorFlow and keras
- A note on using Google Colaboratory
The live seminar module contains:
- A pdf of the presentation
- Link to try Google's cloud vision API
- A page titled Let's try computer vision with images
- Jupyter notebook
- Jupyter notebook with TensorFlow in Google colaboratory (Hello world of course).
- Handwritten digits recognition (the classic machine learning program for beginners).
- Links to web resources (TensorFlow, Keras, and Moments in time)
The post-seminar module contains:
- A Jupyter notebook with a machine learning algorithm
- The annotated Jupyter notebook
- Practice test
- Final test to earn your digital badge