Optimizing PolyACO Training with GPU-Based Parallelization

Tufteland, Torry; Ødesneltvedt, Guro; Goodwin, Morten

doi:10.1007/978-3-319-44427-7_20

Cited by 4 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Parallelisation reduces the runtime by a factor of up to 61.5x, which is a significant improvement compared with the original PolyACO (Tufteland et al 2016). …”

Section: Gpu Performance Comparisonmentioning

confidence: 96%

“…PolyACO is a grid-based polygon algorithm aimed at creating boundaries by surrounding and separating classes guided by ACO Tufteland et al 2016). Figure 2 presents an overview included here for explanatory purposes.…”

Section: Polyacomentioning

confidence: 99%

“…Based on recent advances from Tufteland et al (2016), we now introduce GPU in the training phase. This is done by parallelising the most costly part of the algorithm, the reward function, to the graphics processing unit (GPU) using CUDA (Ryoo et al 2008).…”

Section: Parallelisation With Gpumentioning

confidence: 99%

“…In practice, the GPU is invoked once per ant and returns a two-dimensional array of points in the training set and the number of edges produced by the ant. For more details, we refer the reader to Tufteland et al (2016).…”

Section: Parallelisation With Gpumentioning

confidence: 99%

“…Secondly, some prominent studies, including AntMiner and AntMiner+, have used ACO as a rule-based classifier (Martens et al 2007). Thirdly, some recent advances have introduced PolyACO, a polygon-based classification algorithm Tufteland et al 2016). 1 Interestingly, PolyACO deals with the classification problem in a completely different manner from existing classifiers.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

PolyACO+: a multi-level polygon-based ant colony optimisation classifier

et al. 2017

Self Cite

View full text Add to dashboard Cite

Ant colony optimisation (ACO) for classification has mostly been limited to rulebased approaches where artificial ants walk on datasets in order to extract rules from the trends in the data, and hybrid approaches which attempt to boost the performance of existing classifiers through guided feature reductions or parameter optimisations. A recent notable example that is distinct from the mainstream approaches is PolyACO, which is a proof-ofconcept polygon-based classifier that resorts to ACO as a technique to create multi-edged polygons as class separators. Despite possessing some promise, PolyACO has some significant limitations, most notably, the fact of supporting classification of only two classes, including two features per class. This paper introduces PolyACO+, which is an extension of PolyACO in three significant ways: (1) PolyACO+ supports classifying multiple classes, (2) PolyACO+ supports polygons in multiple dimensions enabling classification with more than two features, and (3) PolyACO+ substantially reduces the training time compared to Poly-ACO by using the concept of multi-levelling. This paper empirically demonstrates that these updates improve the algorithm to such a degree that it becomes comparable to state-of-the-art techniques such as SVM, neural networks, and AntMiner+.

show abstract

“…Parallelisation reduces the runtime by a factor of up to 61.5x, which is a significant improvement compared with the original PolyACO (Tufteland et al 2016). …”

Section: Gpu Performance Comparisonmentioning

confidence: 96%

Section: Polyacomentioning

confidence: 99%