Course Deep Learning Architectures

---

Responsible: Prof. Dr. Beate Rhein

Course

Meets requirements of following modules(MID)

• in active programs
  • Ma TIN2012 DLA

Course Organization

Version created 2013-06-06 VID 1 valid from WS 2012/13 valid to

Course identifiers Long name Deep Learning Architectures CID F07_DLA CEID (exam identifier)

Contact hours per week (SWS) Lecture 2 Exercise (unsplit) 2 Exercise (split) Lab 1 Project Seminar Tutorial(voluntary)

Total contact hours Lecture 30 Exercise (unsplit) 30 Exercise (split) Lab 15 Project Seminar Tutorial (voluntary)

Max. capacity Exercise (unsplit) Exercise (split) Lab 18 Project Seminar

Total effort (hours): 180

Instruction language

• German
• English

Study Level

• Master

Prerequisites

• basic skills in neural networks
• strong programming skills (Python)

Textbooks, Recommended Reading

• I. Sommerville: Software Engineering (Addison-Wesley / Pearson Studium)
• H. Balzert et.al.: Lehrbuch der Softwaretechnik (Spektrum Akademischer Verlag)
  • Basiskonzepte und Req.Eng.
  • Softwaremanagement
• G.E. Thaller: Software- und Systementwicklung (Heise Verlag)
• Bernd Oestereich: Analyse und Design mit UML 2.3 (Oldenbourg)
• Gamma et.al.: Design Patterns, (Addison-Wesley)
• OMG Unified Modeling Language Spec., www.omg.org/uml
• K. Beck: eXtreme Programming (Addison-Wesley Professional)
• Ken Schwaber: Agiles Projektmanagement mit Scrum (Microsoft Press)
• H.D. Litke: Projektmanagement (Hanser)

Instructors

• MSc Jan Bollenbacher

Supporting Scientific Staff

• MSc Jan Bollenbacher

Transcipt Entry

Deep Learning Architectures

Assessment

Type
oE	normal case (except on large numbers of assessments: wE

Total effort [hours]
oE	10

Frequency: 2-3/year

---

Course components

Lecture/Exercise

Objectives

Lerninhalte (Kenntnisse)

• Introduction Machine Learning
• Tooling
  • Handling Machine Learing Frameworks
  • Parallel Computing using GPUs
• Types of Neuronal Networks and application
  • Feedforward Neuronal Networks and Backpropagation
  • Convolutional Networks
  • Recurrent Networks
• Deep Reinforcement Learning
  • Introduction Reinforcement Learning
  • Policy-Gradient Methods
  • Actor-Critic Methods

Fertigkeiten (PFK 2)

• Knowing the mathematic principles of Algorithms
• Using Maschine Learning Frameworks
• Applying Algorithms on problems
  • time series
  • image classification
  • complex environments (Games)

Kompetenzen (PFK 3)

• Classification of problems and identification of feasible solutions
• Evaluation of algorithms
• Independently gaining knowledge in complex field

Additional Component Assessment

Lab

Objectives

Fertigkeiten (PFK 4, PFK 5, PFK 6)

• Transfering knowledge from course on new problems
• Implementing algorithms on new problems
• Scientific Working and experiments
• Interpretation of results

Handlungskompetenz demonstrieren (PFK 7, PSK 1, PSK 3)

• Outline the results in a short paper
• Research in complex field
• Rate paper by relevance
• Rating the experiment results
• Rating the used algorithms

Additional Component Assessment

• Präsentation der Ergebnisse des Praktikums
• Beitrag zum LV-Ergebnis
• Intervall
  • 1/Jahr