Deep learning is the most advanced branch of artificial intelligence today. Inspired by the human brain, it enables computers to learn from examples. Deep learning has been successfully applied to tasks that were previously thought to require human intelligence. To name a few, using deep learning computers can now recognize the contents of images, answer questions, translate from one language to another, and even compose music.

The Deep Learning minor at Inholland University of Applied Sciences gives students a thorough introduction to the field of deep learning. Students learn to build deep learning models using Tensorflow/Keras, a state-of-the-art framework for deep learning in Python. During the minor students will build deep learning networks for various machine learning tasks, such as classification, image recognition and natural language processing.

The minor has a practical orientation and is organized around several programming assignments and a project.  The first part of the minor consists of a series of lectures and lab sessions. Students learn the foundations of deep learning and get hands-on experience with the Tensorflow/Keras framework. Knowledge of deep learning concepts is assessed in a written exam. The second part focusses on the project. In the project the learned concepts are applied to a realistic case from a real organization. In this part students can choose to specialize in either image recognition or natural language processing. Lectures are given about advanced aspects of deep learning related to the selected projects.

After completing the minor the student is able to:

• Explain the working of a deep learning model
• Choose an appropriate deep learning architecture for a real-world problem
• Prepare data sets for deep learning using Python
• Train fully connected, convolutional, and recurrent deep learning models using Tensorflow/Keras
• Conduct experiments to evaluate deep learning models
• Assess the value of a deep learning model for a real-world problem

The minor is open for 3rd and 4th year bachelor students. The following skills are required:

• Python programming
• Basic statistics
• Basic algebra

• Francois Chollet, Deep Learning with Python 2nd edition, Manning Publications.

The Deep Learning minor spans over a period of 5 months, from February 5, 2024 to July 5, 2024.

Credits: 30EC

Contact hours: maximal 3 days per week at school

Part One: Foundations of Deep Learning

Written exam (4EC)
Assignments (8EC)
Literature study and presentation (3EC)

Topics covered in lectures and assignments:

• Introduction to neural networks and deep learning
• Mathematical foundations of deep learning
• Building a basic deep learning model in Tensorflow/Keras
• Evaluating a model using Tensorflow/Keras
• Image recognition using convolutional neural networks
• Using pretrained networks
• Sequence learning using recurrent neural networks
• Text processing
• Generative deep learning

 

Part one is assessed by a number of programming assignments and a written exam. The exam covers the theory behind deep learning. The assignments focus on the use of Tensorflow/Keras for building and evaluating various kinds of deep learning networks. In addition, students do a literature study into an application of deep learning of their choice.

Part Two: Deep Learning in Action

Group project (15EC)

Students choose a project on deep learning for image classification or deep learning for natural language processing. Aspects covered in the project:

• Preparing a data set for deep learning
• Selecting a model architecture and/or pretrained model
• Training a model using Tensorflow/Keras
• Optimizing a model using parameter tuning
• Evaluating the value of a model