Advanced Machine Learning

Deep Learning

Download the training course in PDF format

Registration file

Support for participants with disabilities

The University Life Service – Disability Office provides dedicated support and accommodation services to ensure that participants with disabilities can access training under the best possible conditions. Find out more

Any request for accommodation can be reviewed prior to the start of the training, depending on the specific needs.

For assistance, please contact our disability representative at: sfc-handicap@unistra.fr

Context and Goals of the training program

The aim of this course is to develop an understanding of deep learning and data visualisation. You will gain theoretical knowledge of the different components and architectures of neural networks and apply it to real-world data via supervised and unsupervised approaches. We will use Python and Tensorflow.

Main strenghts:

• A balanced approach combining theoretical foundations with hands-on practice

• Personnalized support and mentoring throughout the program

• Course designed and taught by a recognised expert in the field

Intended participants

This program is intended for professionals who are comfortable working with advanced mathematics, such as computer scientists, physicists, statisticians, engineers, and related profile.

Learning Outcomes

Understand how deep learning algorithms work
Implement and train a deep neural network
Visualise and explore data
Use an existing neural network for new tasks

Program

Day 19:00 – 12:00Theory: Introduction to Deep Learning, Convolutional Neural NetworksPractical Work: Segmentation and classification13:00 – 16:00T: Architectures and cost functionsPW: Regression and classificationDay 29:00 – 12:00T: Advanced training: augmentation and dropoutPW: Segmentation with augmentation13:00 – 16:00T: Transfer learning, pre-trained architecturesPW: Transfer Learning with Deep NeuralDay 39:00 – 12:00T: Dimension reduction and visualisationPW: Eigenfaces13:00 – 16:00T: Stacked, sparse and denoising autoencodersPW: Representation learningDay 1
9:00 – 12:00Theory: Introduction to Deep Learning, Convolutional Neural NetworksPractical Work: Segmentation and classification13:00 – 16:00T: Architectures and cost functionsPW: Regression and classification

Day 2
9:00 – 12:00T: Advanced training: augmentation and dropoutPW: Segmentation with augmentation13:00 – 16:00T: Transfer learning, pre-trained architecturesPW: Transfer Learning with Deep Neural

Day 3
9:00 – 12:00T: Dimension reduction and visualisationPW: Eigenfaces13:00 – 16:00T: Stacked, sparse and denoising autoencodersPW: Representation learning

Access conditions and Prerequisites

Introductory course to Machine Learning or equivalent knowledge: experience in machine learning/data science, programming experience, Python.

If applicable, a telephone interview with the academic director will be conducted to verify that the prerequisites are met.

To know more about the Introductory course, follow this link: https://www.h-ka.de/iww/machine-learning/ueberblick-inhalte.

To register to the Introductory course, you can contact Romina Kolb, director of the Continuing Education Institute of Karlsruhe University of Applied Sciences: romina.kolb@h-ka.de.

Teaching methods and resources

Theoretical input – Practical work, hands-on activities – Slideshows and other course materials – Practical sheets

Academic director

Prof Thomas Lampert, Professor of Computer Science, University of Strasbourg – lampert@unistra.fr

Nature and certification of the training

This training constitutes a skills adaptation and development program.

It leads to the issuance of a certificate of attendance.

An evaluation conducted at the end of the training measures participants’ satisfaction as well as the achievement of the training objectives (knowledge, skills, engagement, and confidence), in accordance with Levels 1 and 2 of the evaluation model developed by Donald Kirkpatrick.