RT Journal Article
T1 Sparse matrix classification on imbalanced datasets using convolutional neural networks
A1 Pichel Campos, Juan Carlos
A1 Pateiro López, Beatriz
K1 Sparse matrix
K1 Classification
K1 Imbalance
K1 Deep learning
K1 CNN
K1 Performance
AB This paper deals with the class imbalance problem in the context of the automatic selectionof the best storage format for a sparse matrix with the aim of maximizing the performance of the sparsematrix vector multiplication (SpMV) on GPUs. Our classi cation method uses convolutional neural networks(CNNs) and proposes several solutions to mitigate the bias toward the majority classes when thedata are not balanced. First, the CNNs are trained using images that represent the sparsity pattern of thematrices, whose pixels are colored according to different matrix features. In addition, we introduce a newnetwork called SpNet, which achieves better results than a standard network as AlexNet in terms of predictionaccuracy even having a more simple architecture. Finally, sampling techniques and cost-sensitive methodshave been studied to give more emphasis on minority classes. The experiments conducted show that ourclassi ers are able to select the best performing format 92.8% of the time, obtaining 98.3% of the maximumattainable SpMV performance.Acomparison to other state-of-the-art classi cation methods is also provided,demonstrating the bene ts of our proposal
PB IEEE
YR 2019
FD 2019
LK http://hdl.handle.net/10347/21079
UL http://hdl.handle.net/10347/21079
LA eng
NO Pichel, J. and Pateiro-Lopez, B., 2019. Sparse Matrix Classification on Imbalanced Datasets Using Convolutional Neural Networks. IEEE Access, 7, 82377-82389
NO This work was supported in part by the Ministerio de Economía y Competitividad (MINECO) under Grant MTM2016-76969-P and Grant RTI2018-093336-B-C21, in part by the Xunta de Galicia under Grant ED431G/08 and Grant ED431C 2018/19, and in part by the European Regional Development Fund (ERDF)
DS Minerva
RD 28 abr 2026