Inhalt

Abstract:

Deep Learning (DL) has attracted much interest in a wide range of applications such as image recognition, speech recognition, and artificial intelligence, both from academia and industry. In this course, you will learn the core elements of neural networks and deep learning, such as convolutional layers, activation and loss functions, and regularization techniques.

Gliederung:

1. Introduction
2. Signal Processing
3. Image Processing
4. Feedforward Networks
5. Loss and Optimization
6. Activations, Convolution and Pooling
7. Regularization
8. Common Practices
9. Architectures
10. Unsupervised Learning
11. Segmentation and Object Detection

Detaillierter Inhalt:

Neural networks have had an enormous impact on research in image and signal processing in recent years. In this course, you will learn all the basics about deep learning in order to understand how neural network systems are built. The course is addressed to students who are new to the field. In the beginning of the course, we introduce you to the topic with some applications of deep learning in the field of medical imaging, digital humanities and industry projects. Before we dive into the core elements of neural networks, there are two lecture units on the fundamentals of signal and image processing to teach you relevant parts of system theory such as convolutions, Fourier transform, and sampling theorem. In the next lecture units, you learn the basic blocks of neural networks, such as backpropagation, fully connected layers, convolutional layers, activation functions, loss functions, optimization, and regularization strategies. Then, we look into common practices for training and evaluating neural networks. The next lecture unit is focusing on common neural network architectures, such as LeNet, Alexnet, and VGG. It follows a lecture unit about unsupervised learning that contains the principles of autoencoders and generative adversarial networks. Lastly, we cover some applications of deep learning in segmentation and object detection.

The accompanying programming exercises will provide a deeper understanding of the workings and architecture of neural networks, in which you will develop a basic neural network from scratch in pure Python without using deep learning frameworks, such as PyTorch or TensorFlow.

At the end of the semester, there will be a written exam.

Lern-/Qualifikationsziele:

-

Lehrveranstaltungstyp:

Virtuelle Vorlesung

Interaktionsformen mit Betreuer/in:

Chat

Interaktionsformen mit Mitlernenden:

Chat

Kursdemo:

zur Kursdemo

Nutzung

Kurs ist konzipiert für:

Friedrich-Alexander-Universität Erlangen-Nürnberg:
Informatik, BSc
Medizintechnik, BSc
Computational Engineering, BSc
Data Science, BSc

TU München:
Biomedical Computing, MSc

Formale Voraussetzungen:

-

Erforderliche Vorkenntnisse:

Basic knowledge of linear algebra and analysis are required (matrix operations, integral summation).
Practical knowledge in the Python programming language is necessary to succeed in the programming exercises.

Hinweise zur Nutzung:

-

Kursumsetzung (verwendete Medien):

-

Erforderliche Technik:

-

Nutzungsentgelte:

für andere Personen als (reguläre) Studenten der vhb Trägerhochschulen nach Maßgabe der Benutzungs- und Entgeltordnung der vhb

Rechte hinsichtlich des Kursmaterials:

-

Verantwortlich

Anbieterhochschule:

Uni Erlangen-Nürnberg (FAU)

Anbieter:

Prof. Dr. Andreas Maier

Autoren:

Andreas Maier

Betreuer:

Deep Learning Tutorenteam

Prüfung