Keras TensorFlow Course for machine learning

TensorFlowLogo In the last ten years there has been a revolution in the field of Artificial Intelligence, mainly as the result of the revival of an approach called “Deep Learning”. For instance, AlphaGo and ChatGPT. Deep Learning is the subfield of Machine Learning which is based on Artificial Neural Networks with many “layers”.

The Deep-Learning course discusses the main ideas and techniques of Deep Learning and explains the practical implementation of Deep Learning algorithms using Python and Keras in detail. The programming part in this course is about 50% of the time.

At the end of the course participants will be able to program state of the art algorithms in computer vision with a few lines of code.

The course assumes familiarity with core Python. It is an advantage to have basic knowledge of linear algebra and calculus, but not a must. The topics covered are:

1) Introduction to neural networks

  • Design of neural networks.
  • Main applications of deep neural networks.
  • Overview of frameworks for neural networks.

2) The Sequential Model in Keras

  • Setting the TensorFlow logging level.
  • Backpropagation.
  • Activation functions.
  • Loss functions.
  • Optimizers.
  • Convolutional networks.
  • Important Keras Layers for image input.
  • Generally important Keras Layers.
  • Obtaining information about Keras Layer shapes.
  • A Keras Model can act as a Keras Layer.
  • Logging and monitoring of the methos.
  • Monitor progress of with TensorBoard.

3) The Keras functional API

  • Programming complex network architectures.
  • Shared Layers.
  • String- or object arguments for the compile() method.

4) Transfer learning. Reuse a pretrained model.

  • Save and load a model.
  • Loading pretrained models from the internet.
  • Modifying pretrained models
  • Fixing weights in certain Layers.
  • Remove layers from a model.
  • Models with several input and/or output layers.

5) Problematic situations and their remedies

  • Overfitting.
  • Exploding / vanishing gradients.
  • Very large data sets.

The standard course duration is 2 days. On request, this course can be combined with one of the other courses with a duration between 2 and 5 days. Very popular is the combination with the Scikit-Learn-course. The course size is between 1 and 9 participants.

If you are interested in this course, please send us a message, since we plan courses dynamically on demand. (Price list).