Cooperative Coevolutionary Approximation in HOG-based Human Detection Embedded System

Wiglasz, Michal; Sekanina, Lukáš

doi:10.1109/ssci.2018.8628910

Cited by 5 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By using simple functions such as addition, minimum and maximum, CGP provides approximate implementations of arctan and square root functions [44]. These evolved functions then replace their standard implementations in the gradient orientation and magnitude computation modules that are critical components of the gradient histogram computation of the histogram of oriented gradients (HOG) feature extraction method.…”

Section: Higher Levels Of Abstractionmentioning

confidence: 99%

Evolutionary Algorithms in Approximate Computing: A Survey

Sekanina

2021

JICS

Self Cite

View full text Add to dashboard Cite

In recent years, many design automation methods have been developed to routinely create approximate implementations of circuits and programs that show excellent trade-offs between the quality of output and required resources. This paper deals with evolutionary approximation as one of the popular approximation methods. The paper provides the first survey of evolutionary algorithm (EA)-based approaches applied in the context of approximate computing. The survey reveals that EAs are primarily applied as multi-objective optimizers. We propose to divide these approaches into two main classes: (i) parameter optimization in which the EA optimizes a vector of system parameters, and (ii) synthesis and optimization in which EA is responsible for determining the architecture and parameters of the resulting system. The evolutionary approximation has been applied at all levels of design abstraction and in many different applications. The neural architecture search enabling the automated hardware-aware design of approximate deep neural networks was identified as a newly emerging topic in this area.

show abstract

Section: Higher Levels Of Abstractionmentioning

confidence: 99%

Evolutionary Algorithms in Approximate Computing: A Survey

Sekanina

2021

JICS

Self Cite

View full text Add to dashboard Cite

show abstract

“…As computing becomes ubiquitous, with the internet-of-things and in particular mote computing, there will be a demand to run increasingly sophisticated algorithms (such as computer vision and machine learning) which may require vector normalisation on non-standard hardware lacking support for some mathematics functions and basic facilities we usually take for granted (such as a power supply). Hence there may be a demand for non-conventional mathematics implementations possibly providing unconventional trade-offs with energy consumption [40] or accuracy [59,64] and a software based solution may be preferred to hardware circuits, such as [21].…”

Section: Testing the Evolved Log2 Functionmentioning

confidence: 99%

Genetic Improvement of Data for Maths Functions

Langdon

Krauss

2021

ACM Trans. Evol. Learn. Optim.

View full text Add to dashboard Cite

We use continuous optimisation and manual code changes to evolve up to 1024 Newton-Raphson numerical values embedded in an open source GNU C library glibc square root sqrt to implement a double precision cube root routine cbrt, binary logarithm log2 and reciprocal square root function for C in seconds. The GI inverted square root x -1/2 is far more accurate than Quake’s InvSqrt, Quare root. GI shows potential for automatically creating mobile or low resource mote smart dust bespoke custom mathematical libraries with new functionality.

show abstract

“…As computing becomes ubiquitous, with the internet-of-things and in particular mote computing, there will be a demand to run increasingly sophisticated algorithms (such as computer vision and machine learning) which may require vector normalisation on nonstandard hardware lacking support for some maths functions and basic facilities we usually take for granted (such as a power supply). Hence there may be a demand for non-conventional maths implementations possibly providing unconventional trade-offs with energy consumption [18] or accuracy [25,29] and a software based solution may be preferred to hardware circuits, such as [7].…”

Section: Quakementioning

confidence: 99%