Deep Learning (DL)5 ECTS
(englische Bezeichnung: Deep Learning)

Lehrveranstaltungen:

• • Deep Learning
    (Vorlesung, 2 SWS, Andreas Maier, Mi, 8:30 - 10:00, 0.68)
  • Deep Learning Exercises
    (Übung, 2 SWS, Tobias Würfl et al., Mo, 16:00 - 18:00, 0.01-142)

Empfohlene Voraussetzungen:

Es wird empfohlen, folgende Module zu absolvieren, bevor dieses Modul belegt wird:

Pattern Recognition Deluxe (WS 2016/2017)
Introduction to Pattern Recognition Deluxe (WS 2016/2017)

Inhalt:

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. This lecture introduces the core elements of neural networks and deep learning, it comprises:

• (multilayer) perceptron, backpropagation, fully connected neural networks

• loss functions and optimization strategies

• convolutional neural networks (CNNs)

• activation functions

• regularization strategies

• common practices for training and evaluating neural networks

• visualization of networks and results

• common architectures, such as LeNet, Alexnet, VGG, GoogleNet

• recurrent neural networks (RNN, TBPTT, LSTM, GRU)

• deep reinforcement learning

• unsupervised learning (autoencoder, RBM, DBM, VAE)

• generative adversarial networks (GANs)

• weakly supervised learning

• applications of deep learning (segmentation, object detection, speech recognition, ...)
  

The accompanying exercises will provide a deeper understanding of the workings and architecture of neural networks.

Lernziele und Kompetenzen:

The students

• explain the different neural network components,

• compare and analyze methods for optimization and regularization of neural networks,

• compare and analyze different CNN architectures,

• explain deep learning techniques for unsupervised / semi-supervised and weakly supervised learning,

• explain deep reinforcement learning,

• explain different deep learning applications,

• implement the presented methods in Python,

• autonomously design deep learning techniques and prototypically implement them,

• effectively investigate raw data, intermediate results and results of Deep Learning techniques on a computer,

• autonomously supplement the mathematical foundations of the presented methods by self-guided study of the literature,

• discuss the social impact of applications of deep learning applications.
  

Literatur:

• Ian Goodfellow, Yoshua Bengio, Aaron Courville: Deep Learning. MIT Press, 2016.

• Christopher Bishop: Pattern Recognition and Machine Learning, Springer Verlag, Heidelberg, 2006

• Yann LeCun, Yoshua Bengio, Geoffrey Hinton: Deep learning. Nature 521, 436–444 (28 May 2015)

Weitere Informationen:

Verwendbarkeit des Moduls / Einpassung in den Musterstudienplan:
Das Modul ist im Kontext der folgenden Studienfächer/Vertiefungsrichtungen verwendbar:

1. Informatik (Master of Science)
   (Po-Vers. 2010 | TechFak | Informatik (Master of Science) | Wahlpflichtbereich | Säule der anwendungsorientierten Vertiefungsrichtungen | Vertiefungsrichtung Mustererkennung)
2. Medizintechnik (Master of Science)
   (Po-Vers. 2013 | TechFak | Medizintechnik (Master of Science) | Studienrichtung Medizinische Bild- und Datenverarbeitung | M2 Ingenieurwissenschaftliche Kernmodule (BDV))

Studien-/Prüfungsleistungen:

Deep Learning (Prüfungsnummer: 901895)

(englischer Titel: Deep Learning)

Prüfungsleistung, mündliche Prüfung, Dauer (in Minuten): 30, benotet

Anteil an der Berechnung der Modulnote: 100.0 %

Prüfungssprache: Deutsch oder Englisch

Erstablegung: WS 2017/2018, 1. Wdh.: SS 2018

1. Prüfer:

Andreas Maier