RTDL 2017: IEEE Swiss Circuits and Systems “Real-time deep learning” Workshop

Many thanks to all the great presenters of talks, posters and demos and to the lively participants.

Photos of the event are available here.

Reason for workshop: Autonomous driving and flying, mobile devices, domotics, and IoT are rapidly driving the development of both deep learning models and quick, power-efficient deep neural network inference accelerators. At the same time, theory advances are enabling potential incremental, online and reinforcement learning methods that might be computed on devices rather than in the cloud. Come to this workshop to hear about the latest results from these real-time deep neural networks. The workshop includes talks, poster/demo sessions and apero to promote lively discussions and networking in this fast-growing field.

Background: This workshop is one of a series organized by Swiss CAS/ED, including the 2011 “Swiss Image and Vision Sensors” and the 2014 “Memristive Devices and Neuromorphic Applications” workshops, along with several student workshops and a series of IEEE CAS/ED distinguished lectures.

Organizers: IEEE Swiss Chapter of Circuits and Systems and Electron Devices Society
Date: Tuesday 05.12.2017, 9:30 to ~19:00

Place: UZH Irchel Campus, Room Y15 G19; demo, poster and apero area in corridor between Y15 G19 and building Y55. (See map at the bottom of the page.)

Program

Talks are all strictly 15m talk + 5m discussion.
See
here for registered demos and posters.
See
here for RTDL 2017  talk abstracts.

Final Schedule. All presentations will be in English.

Time

Speaker

Organization

Title (see here for abstracts)

9:30

Coffee, poster/demo setup, in hallway on Floor F between building 19 and 55

10:00

Welcome and IEEE Swiss CAS/ED news

10:20

Kevis-Kokitsi Maninis

Computer Vision Lab (CVL, Luc van Gool), ETH Zurich, D-ITET

Real-time segmentation using deep learning for medical imaging

10:40

Torsten Sattler

Computer Vision and Geometry Lab (CVG, Marc Pollefeys), ETH Zurich, D-INFK

Learning Towards Real-Time Stereo Inference

11:00

Alessandro Giusti

IDSIA (Luca Gambardella)

Deep learning applications to mobile robotics: navigating natural environments and interacting with people

11:20

Coffee Break (20m)

11:40

Tobi Delbruck

Sensors Group, Inst. of Neuroinformatics, UZH-ETH Zurich (D-ITET)

Exploiting spatial sparsity and event cameras for data driven CNN acceleration

12:00

Shih-Chii Liu

Sensors Group, Inst. of Neuroinformatics, UZH-ETH Zurich (D-ITET)

Exploiting Input Timing Information and Temporal Sparsity in Recurrent Deep Networks

12:20

Standing Lunch and Posters (1h40)

14:00

Abu Sebastian

IBM Research, Memory and Cognitive Technologies group

Computational memory: Towards non-von Neumann accelerators

14:20

Luca Benini

Inst. of Integrated Systems (IIS, ETH Zurich, D-ITET)

Micropower deep Learning: toward pj/OP and beyond

14:40

Short break (10m)

14:50

Otmar Hilliges

Advanced Interactive Technologies Lab (AIT, Inst. of Visual Computing, ETH-Zurich D-INFK)

Deep-Learning for Intelligent Human Machine Interaction

15:10

Short break (10m) setup for special event

15:20

Special event (20m) - Machine Learning Jeopardy. A battle of deep knowledge of machine learning facts and trivia between teams of “experts”

15:40

Group photo (5m) - front of lecture hall

15:45-18:00

Poster, Demonstrations, and Happy Hour in Corridor

Registration 

  1. All participants must register and pay for the workshop even if not showing a demo or poster. Registrations are now closed.
  1. IEEE members: You will need your IEEE membership number for discount. 
  2. Your registration will not be valid until you select either paypal (and pay) or invoice. Your registration will not be complete until you complete one of these options.
  3. Cutoff date for registration is Friday 24 Nov 2017. Please contact Shih-Chii Liu for questions about registration after this date.
  4. Registration limit: 100 persons; first come, first serve.
  1. Poster/demo registration is now closed.
  1. Posters should be A0 format, portrait type (vertical)
  2. Demos also get a poster board
  3. We have space for only 20 posters and 6 demos; first come, first serve. 
    Current status:

Num posters

19

Num demos

3

  1. Poster/Demo registration deadline: Friday 24 Nov 2017.
  2. Check spreadsheet of posters/demos for your demo, if you selected that it be listed in the program.

Registration cost: 80/60 CHF for postdocs and above, 40/30 CHF for phd students and below (non-IEEE/IEEE member). Lunch and Happy Hour apero included.

Sponsors

Organzation and Contact

How to get to workshop by tram

See here for a map of UZH Irchel Campus. To get to workshop, take tram to either Irchel or Milchbuck tram stops and follow UZH signs to Room Y15 G19 (Y15 is the building and G means level G).

Poster/demo area (in hallway connecting building 19 and 55)