Extending parallelization of the self-organizing map by combining data and network partitioned methods

Richardson, Trevor; Winer, Eliot H.

doi:10.1016/j.advengsoft.2015.05.003

Cited by 9 publications

(3 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [29], a SOM implementation that combines both data and model parallelism is described. This implementation pushes the parallelization capability of the batch training algorithm to the extreme, as not only the training set is split in chunks to be processed independently by copies of the map (data parallelism) but each copy is also partitioned at unit weightslevel, so that a separated GPU thread handles the updates of a single dimension of each neuron.…”

Section: Related Workmentioning

confidence: 99%

XPySom: High-Performance Self-Organizing Maps

Mancini

Ritacco

Lanciano

et al. 2020

2020 IEEE 32nd International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD)

View full text Add to dashboard Cite

In this paper, we introduce XPySom, a new opensource Python implementation of the well-known Self-Organizing Maps (SOM) technique. It is designed to achieve high performance on a single node, exploiting widely available Python libraries for vector processing on multi-core CPUs and GP-GPUs. We present results from an extensive experimental evaluation of XPySom in comparison to widely used open-source SOM implementations, showing that it outperforms the other available alternatives. Indeed, our experimentation carried out using the Extended MNIST open data set shows a speed-up of about 7x and 100x when compared to the best open-source multi-core implementations we could find with multi-core and GP-GPU acceleration, respectively, achieving the same accuracy levels in terms of quantization error.

show abstract

Section: Related Workmentioning

confidence: 99%

XPySom: High-Performance Self-Organizing Maps

Mancini

Ritacco

Lanciano

et al. 2020

2020 IEEE 32nd International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD)

View full text Add to dashboard Cite

show abstract

“…Another parallel SOM architecture [10,11] were implemented by dividing the calculation into two kernels, the first kernel calculates the Euclidean distance and finds the BMU and the second kernel updates the neighbor weights. Yet another architecture works were using one kernel to calculate Euclidean distance and find BMU, while neighbor weight update was performed on the CPU [12].…”

Section: Previous Researchmentioning

confidence: 99%

“…Yet another architecture works were using one kernel to calculate Euclidean distance and find BMU, while neighbor weight update was performed on the CPU [12]. Experiments to reduce computational time are also conducted with a combination of common methods used in SOM, such as network partition and data partition methods that are run in parallel [11]. Another method that is also utilized is parallel reduction as a solution to reduce computational time [13].…”

Section: Previous Researchmentioning

confidence: 99%

Enhancing Performance of Parallel Self-Organizing Map on Large Dataset with Dynamic Parallel and Hyper-Q

Sibero

Sitompul

Nasution

2018

jocai

View full text Add to dashboard Cite

Self-Organizing Map (SOM) is an unsupervised artificial neural network algorithm. Even though this algorithm is known to be an appealing clustering method,many efforts to improve its performance are still pursued in various research works. In order to gain faster computation time, for instance, running SOM in parallel had been focused in many previous research works. Utilization of the Graphics Processing Unit (GPU) as a parallel calculation engine is also continuously improved. However, total computation time in parallel SOM is still not optimal on processing large dataset. In this research, we propose a combination of Dynamic Parallel and Hyper-Q to further improve the performance of parallel SOM in terms of faster computing time. Dynamic Parallel and Hyper-Q are utilized on the process of calculating distance and searching best-matching unit (BMU), while updating weight and its neighbors are performed using Hyper-Q only. Result of this study indicates an increase in SOM parallel performance up to two times faster compared to those without using Dynamic Parallel and Hyper-Q.

show abstract

Adapting Self-Organizing Map Algorithm to Sparse Data

Melka¹,

Mariage²

2019

Studies in Computational Intelligence

View full text Add to dashboard Cite

Extending parallelization of the self-organizing map by combining data and network partitioned methods

Cited by 9 publications

References 26 publications

XPySom: High-Performance Self-Organizing Maps

XPySom: High-Performance Self-Organizing Maps

Enhancing Performance of Parallel Self-Organizing Map on Large Dataset with Dynamic Parallel and Hyper-Q

Adapting Self-Organizing Map Algorithm to Sparse Data

Contact Info

Product

Resources

About