Advanced Topics in Communications Electronics

Vortragende/r (Mitwirkende/r)
Nummer0820675069
Art
Umfang3 SWS
SemesterWintersemester 2021/22
UnterrichtsspracheEnglisch
Stellung in StudienplänenSiehe TUMonline

Teilnahmekriterien

Lernziele

The primary goal of this course is for you to acquire fundamental knowledge of all aspects in artificial intelligence and at the same time experience designing a complex neural network on a resource-limited embedded platform and all of the various tradeoffs/considerations that go along with such design work. The secondary goals of the course are to focus and hone your writing and oral communication skills. By the end of the course, you should be able to assist as a system designer well equipped with knowledge of artificial intelligence and embedded system programming.

Beschreibung

This course covers the fundamental of artificial intelligence for embedded systems including but not limited to deep learning, and advanced topics such as AutoML, graph neural networks, federated/distributed learning and security. This course will also cover various deep compression techniques for them to be deployed on embedded devices for mobile or edge applications. The course will include comprehensive team-based hand-on design experience of a neural network on an embedded platform for a real application of choice. Projects involve neural network design, training, compression, prototype implementation, and documentation. Group project management skills, including scheduling and project tracking are stressed. Instructor: Prof. Yiyu Shi, Professor, Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, IN, 46556 USA Email: yshi4@nd.edu Course Objectives: • Understand common machine learning techniques and the rationale/mechanism behind them • Understand the strengths and limitations of different machine learning algorithms and how to address them • Design and implement proper machine learning algorithms on a target embedded system. • Understand the practical issues related to the scalability, security and trustworthiness of machine learning algorithms on the edge • Understand the state-of-the-art machine learning development and applications in academia and industry • Deliver quality technical reports and oral presentations. • Work effectively in a team and know how to use effective meeting management skills. Course Policies: 1. Lecture • Not all content discussed in class will be on-line so taking notes during class is highly encouraged. In short, anything that is written down on the board, you can write down as well. • Attendance at lectures is required. If you must miss a lecture, please contact the instructor in advance. • Lectures will be driven by student interaction in addition to the standard lecture material. 2. Design project • The design project will be a semester long project • The design project team will consist of 4 students as approved by the instructor. Teaming is strictly required. • The design project will have several deliverables as noted with the grades. 3. Homework / Labs • There are no graded homework or labs. Optional enrichment activities will be provided but are not required. 4. Teamwork • The teams will be evaluated twice during the semester. • Teamwork does not mean that one person does the work one week and the other person does the work the next week. If you sign your name to the report, it means that you have participated in solving the problems. There is no mid-term nor is there a final exam for the class. Topics to be covered (Tentative; content and order subject to change) Introduction of Machine Learning: An Overview (1 lecture) Data Preprocessing (1 lecture) Clustering and Decision Trees (1 lecture) Support Vector Machine (1 lecture) Neural Networks (2 lectures) Deep Learning (2 lectures) Neural Network Compression (3 lectures) AutoML (2 lectures) Machine Learning Security (2 lectures) Federated/Decentralized Learning (1 lecture) Natural Language Processing (1 lecture) Graph Neural Networks (1 lecture) Generative Adversarial Networks (1 lecture) State-of-the-Art Advances in Industry and Academia (various)

Studien-, Prüfungsleistung

Grading: Class Attendance 10% Class Presentation/Discussion 10% Project Proposal 10% Nov. 12 Status Report 1 10% Dec. 17 Status Report 2 10% Jan. 21 Final Presentation 20% Feb. 11 Final Report 30% Feb. 11 There are no late submissions. Any extensions must require extenuating circumstances or a priori negotiation. In short, be pro-active, not reactive. Honor Code: You have to develop your own project. If you want to use anything you find online, please consult the instructor first.

Empfohlene Literatur

None (Course materials/videos designed by the instructor will be provided.)

Links


Vollständiges Lehrangebot

Bachelorbereich: BSc-EI, MSE, BSEEIT

 

WS

SS

Diskrete Mathematik für Ingenieure (BSEI, EI00460)

Discrete Mathematics for Engineers (BSEEIT) (Schlichtmann) (Januar)

 

P/P

WS

SS

Entwurf digitaler Systeme mit VHDL u. System C (BSEI, EI0690) (Ecker)

P/m

 

SS

Entwurfsverfahren für integrierte Schaltungen (MSE, EI43811) (Schlichtmann)

P/P

WS

 

Methoden der Unternehmensführung (BSEI, EI0481) (Weigel)

-/P

WS

 

Praktikum System- und Schaltungstechnik (BSEI, EI0664) (Schlichtmann et al.)

--

 

SS

Schaltungssimulation (BSEI, EI06691) (Gräb/Schlichtmann)

P/P

 

Masterbereich: MSc-EI, MSCE, ICD

 

SS

Advanced Topics in Communication Electronics (MSCE, MSEI, EI79002)

 

WS

 

Electronic Design Automation (MSCE, MSEI, EI70610) (B. Li, Tseng)

-/?

WS

 

Design Methodology and Automation (ICD) (Schlichtmann) (Nov)

--

WS

SS

Machine Learning: Methods and Tools (MSCE, MSEI, EI71040) (Ecker)

O/O

WS

SS

SS

Mathematical Methods of Circuit Design (MSCE, MSEI, EI74042) (Gräb)

Simulation and Optimization of Analog Circuits (ICD) (Gräb) (Mai)

P/P

--

WS

 

Mixed Integer Programming and Graph Algorithms in Engineering Problems (MSCE, MSEI, EI71059) (Tseng)

-/Om

WS

SS

Numerische Methoden der Elektrotechnik (MSEI, EI70440) (Schlichtmann oder Gräb)

P/P

WS

WS

SS

Seminar VLSI-Entwurfsverfahren (MSEI, EI7750) (Schlichtmann/Müller-Gritschneder)

Seminar on Topics in Electronic Design Automation (MSCE, EI77502) (Schlichtmann/Müller-Gritschneder)

P/P

P/P

WS

SS

Synthesis of Digital Systems (MSCE, MSEI, EI70640) (Müller-Gritschneder)

P/P

WS

 

Testing Digital Circuits (MSCE, MSEI, EI50141) (Otterstedt)

-/?

WS

 

Timing of Digital Circuits (MSCE, MSEI, EI70550) (B. Li, Zhang)

-/O

WS

SS

VHDL System Design Laboratory (MSCE, MSEI, EI7403) (Schlichtmann)

O/O

 

Die Spalte ganz rechts bezeichnet die Form der Vorlesung/Prüfung im SS 2022. O=online, P=physische Präsenz, m=mündlich, Präsenz oder online. Version: 08.02.2022

The column on the very right denotes the type of course/exam in SS 2022. O=online, P=physical presence, m=oral, presence or online. Version: February 8, 2022

 

MSE: Munich School of Engineering (TUM)

BSEEIT: Bachelor in Electrical Engineering and Information Technology (TUM-Asia)

ICD: Master of Science in Integrated Circuit Design (TUM-Asia)

MSCE: Master of Science in Communications Engineering (TUM)

MSEI: Master of Science in Elektrotechnik und Informationstechnik

BSEI: Bachelor of Science in Elektrotechnik und Informationstechnik

 

Aktuelle Infos zur Lehre/Current information on teaching: https://www.tum.de/die-tum/aktuelles/coronavirus/studium/, www.ei.tum.de