Approximate LSTMs for time-constrained inference: Enabling fast reaction in self-driving cars
File(s)1905.00689v1.pdf (1.09 MB)
Working paper
Author(s)
Kouris, Alexandros
Venieris, Stylianos I
Rizakis, Michail
Bouganis, Christos-Savvas
Type
Working Paper
Abstract
The need to recognise long-term dependencies in sequential data such as video
streams has made LSTMs a prominent AI model for many emerging applications.
However, the high computational and memory demands of LSTMs introduce
challenges in their deployment on latency-critical systems such as self-driving
cars which are equipped with limited computational resources on-board. In this
paper, we introduce an approximate computing scheme combining model pruning and
computation restructuring to obtain a high-accuracy approximation of the result
in early stages of the computation. Our experiments demonstrate that using the
proposed methodology, mission-critical systems responsible for autonomous
navigation and collision avoidance are able to make informed decisions based on
approximate calculations within the available time budget, meeting their
specifications on safety and robustness.
streams has made LSTMs a prominent AI model for many emerging applications.
However, the high computational and memory demands of LSTMs introduce
challenges in their deployment on latency-critical systems such as self-driving
cars which are equipped with limited computational resources on-board. In this
paper, we introduce an approximate computing scheme combining model pruning and
computation restructuring to obtain a high-accuracy approximation of the result
in early stages of the computation. Our experiments demonstrate that using the
proposed methodology, mission-critical systems responsible for autonomous
navigation and collision avoidance are able to make informed decisions based on
approximate calculations within the available time budget, meeting their
specifications on safety and robustness.
Date Issued
2019-05-02
Citation
2019
Publisher
arXiv
Copyright Statement
© 2019 The Authors
Identifier
http://arxiv.org/abs/1905.00689v1
Subjects
eess.SP
eess.SP
cs.LG
cs.RO
Publication Status
Published