Competitive Problem Solving with Deep Learning (Sommersemester 2018)

Dozent: Dr. Haojin Yang (Internet-Technologien und -Systeme)
Tutoren: Joseph Bethge Goncalo Filipe Torcato Mordido

Allgemeine Information

• Semesterwochenstunden: 4
• ECTS: 6
• Benotet: Ja
• Einschreibefrist: 20.04.2018
• Lehrform: Vorlesung / Projekt
• Belegungsart: Wahlpflichtmodul
• Maximale Teilnehmerzahl: 50

Studiengänge, Modulgruppen & Module

Beschreibung

Artificial intelligence (AI) is the intelligence exhibited by computer. This term is applied when a machine mimics "cognitive" functions that humans associate with other human minds, such as "learning" and "problem solving". From the year 2006 "Deep Learning" (DL) has attracted more and more attentions in both academia and industry. Deep learning or deep neural networks is a branch of machine learning based on a set of algorithms that attempt to learn representations of data and model their high level abstractions. Recently DL gives us break-record results in many novel areas as e.g., beating human in strategic game systems like Go (Google’s AlphaGo), self-driving cars, achieving dermatologist-level classification of skin cancer etc.

In this lecture, we will study both theoretical foundations and practical use cases of deep learning techniques. Moreover, a subsequent competitive project will let you get hands-on experience.

The outline of the lecture is thus as follows:

Lecture part:

• Theoretical Foundation   
  • Machine learning foundation
  • Neural Network I
  • Neural Network II
  • Architectures design and advanced techniques in Deep Learning
  • Deep Learning frameworks, understanding and visualization, latest research topics
• Practical use case
  • Build neural network from scratch
  • Advanced techniques in practice
  • Deep learning for computer vision problems
  • Generative networks

Projekt part:

An Image Recognition Challenge

The participants of the course will freely group into several teams (the number of members of each team should be balanced) that try to solve a challenging image recognition problem. The teaching team is responsible for data preparation, prize preparation, organization, evaluation, and be ready for discussion with each team.

Course language: German and English

Literatur

• Ian J. Goodfellow and Yoshua Bengio and Aaron Courville, "Deep Learning", online version
• Pedro Domingos “The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World”
• Christopher M. Bishop “Pattern Recognition and Machine Learning” google it
• cs231n tutorials: Convolutional Neural Networks for Visual Recognition
• Deep Learning courses at Coursera
• "Deep Learning - The Straight Dope” deep learning tutorials created by MXNet team
• Caffe: Deep learning framework by the BVLC
• Chainer: A flexible framework of neural networks
• MXNet: A Flexible and Efficient Library for Deep Learning
• Tensorflow: An open-source machine learning framework

Leistungserfassung

The final evaluation will be based on:

• Written exam 40%

• Final presentation 20%

• Report/Documentation and Implementation 40%

• Participation in the lecture (bonus points)

Termine

Montag, 13.30-15.00

Room: HS 3 not in H-2.57 anymore (except 23.04.2018 and 18.06.2018 in HS 1)

Lecture part:

09.04.2018	Course Introduction (PDF)
bis 20.04.2018	Belegungsfrist/Prüfungsanmeldungsfrist
16.04.2018	Machine learning foundation (T) (PDF) (codes)
23.04.2018	Neural Network I (T) (in HS 1) (PDF) (codes)
30.04.2018	Build network from scratch (P) (PDF)(codes)
07.05.2018	Neural Network II (T) (PDF)
14.05.2018	Advanced techniques in practice (P) (PDF) (codes) (MP4)
28.05.2018	CNN architectures and advanced techniques in Deep Learning (T) (PDF) (caffe prototxt files for Netscope tool) (Netscope)
04.06.2018	Deep learning for computer vision problems (P) (PDF) (code)
11.06.2018	Deep Learning visualization techniques, Competition Introduction (T) (PDF)
18.06.2018	Generative networks (P) (in HS 1) (PDF)
25.06.2018	Written exam (in HS 1)

(T: theoretical lecture, P: practical use cases)

Projekt part:

11.06.2018	Challenge open: release training und validation data, grouping
Weekly	Weekly individual meeting with each team
02.07.2018	Release test set1
09.07.2018	Release pre-ranking result based on the test result offered by participants
10.07.2018	Model submission: Tutors will run the models using a secret test dataset
16.07.2018	Final Presentation (all PDFs), release final ranking result, awards granting
bis 31. August	Implementation + Paper submission (LaTeX template)

Zurück