Workshop Program

08:15 - 10:00    HiPEAC Conference Keynote

10:00 - 11:00    Welcome

Invited Talk
Andreas Moshovos, University of Toronto
A Bit-Pragmatic Approach to Accelerating Convolutional Neural Networks for Deep Learning
[Abstract][Slides]

11:00 - 11:30    Coffee Break

11:30 - 13:00    Session 1: Approximation at the Hardware and Software Levels

Session Chair: George Karakonstantis (Queen's University Belfast)

11:30 - 11:50
Vassilis Vassiliadis, Konstantinos Parasyris, Christos D. Antonopoulos, Spyros Lalis and Nikolaos Bellas. Using artificial neural networks for error detection in unreliable computations.
[Abstract][Slides][Paper]

11:50 - 12:10
Mario Barbareschi, Antonino Mazzeo, Domenico Amelino, Alberto Bosio and Antonio Tammaro. Implementing Approximate Computing Techniques by Automatic Code Mutation.
[Abstract][Slides][Paper]

12:10 - 12:30
Aurangzeb and Rudolf Eigenmann. PROCsimate: A Scheme for Approximating Procedures with Dynamic Quality Monitoring and Result Guarantees.
[Abstract][Slides]>][Paper]

12:30 - 12:45
Igor Neri, Miquel López-Suárez and Luca Gammaitoni. Thermodynamic limits for approximate MEMS memory devices (Short paper).
[Abstract][Slides][Paper]

12:50 - 14:00    Lunch Break

14:00 - 15:30    Session 2: Applications of Approximate Computing and Modelling

Session Chair: Christos D. Antonopoulos (University of Thessaly)

14:00 - 14:20
Imran Wali, Marcello Traiola, Arnaud Virazel, Patrick Girard, Mario Barbareschi and Alberto Bosio. Can we Approximate the Test of Integrated Circuits?
[Abstract][Slides][Paper]

14:20 - 14:40
Oscar Palomar, Andy Nisbet, John Mawer, Graham Riley and Mikel Luján. Reduced precision applicability and trade-offs for SLAM algorithms.
[Abstract][Slides]>][Paper]

14:40 - 15:00
Jens Deussen and Uwe Naumann. Compression of Higher Derivative Tensors in Stochastic Significance Analysis.
[Abstract][Slides]>][Paper]