The self organising map is a machine learning algorithm used to produce low dimensional representations of high dimensional data. While the process is becoming more and more useful with the rise of big data, it is hindered by the sheer amount of time the algorithm takes to run serially. This project produces a parallel version […]

August 11, 2015 by hgpu

The work presented in this PhD studies and proposes cellular computation parallel models able to address different types of NP-hard optimization problems defined in the Euclidean space, and their implementation on the Graphics Processing Unit (GPU) platform. The goal is to allow both dealing with large size problems and provide substantial acceleration factors by massive […]

April 27, 2014 by hgpu

## MAP-based Brain Tissue Segmentation using Manifold Learning and Hierarchical Max-Flow regularization

We developed a fully-automatic multi-atlas initialized segmentation algorithm for tissue segmentation using multi-sequence MR images. The Generalized Hierarchical Max-Flow (HMF) [1] framework proposed in [2] is employed to regularize a maximum a-posteriori data term with a linear label-ordering constraint [3]. The data term is derived from two probabilistic cost functions, i) an intensity model from […]

April 12, 2014 by hgpu

Scientific computations have been using GPU-enabled computers successfully, often relying on distributed nodes to overcome the limitations of device memory. Only a handful of text mining applications benefit from such infrastructure. Since the initial steps of text mining are typically data intensive, and the ease of deployment of algorithms is an important factor in developing […]

August 31, 2013 by hgpu

Somoclu is a C++ tool for training self-organizing maps on large data sets using a high-performance cluster. It builds on MPI for distributing the workload across the nodes of the cluster. It is also able to boost training by using CUDA if graphics processing units are available. A sparse kernel is included, which is useful […]

May 8, 2013 by hgpu

In this paper we introduce a MapReduce-based implementation of self-organizing maps that performs compute-bound operations on distributed GPUs. The kernels are optimized to ensure coalesced memory access and effective use of shared memory. We have performed extensive tests of our algorithms on a cluster of eight nodes with two NVidia Tesla M2050 attached to each, […]

May 10, 2012 by hgpu

In this study, we visualize Pareto-optimum solutions derived from multiple-objective optimization using spherical self-organizing maps (SOMs) that lay out SOM data in three dimensions. There have been a wide range of studies involving plane SOMs where Pareto-optimal solutions are mapped to a plane. However, plane SOMs have an issue that similar data differing in a […]

April 5, 2012 by hgpu

We evaluate a novel implementation of a Self-Organizing Map (SOM) on a Graphics Processing Unit (GPU) cluster. Using various combinations of OpenCL, CUDA, and two different graphics cards, we demonstrate the scalability of the SOM implementation on one to eight GPUs. Results indicate that while the algorithm scales well with the number of training samples […]

February 11, 2012 by hgpu

This paper describes a utilization of the Self Organizing Map (SOM) method for the analysis of power outage data. SOM, to be already used in many fields, is based on the Kohonen self-organizing neural network and it is known to capture underlying concepts. We apply this method for a unified database of power outages to […]

February 5, 2012 by hgpu

Image recognition systems require large image data sets for the training process. The annotation of such data sets through users requires a lot of time and effort, and thereby presents the bottleneck in the development of recognition systems. In order to simplify the creation of image recognition systems it is necessary to develop interaction concepts […]

December 22, 2011 by hgpu

The Self-Organizing Maps (SOMs) are popular artificial neural networks that are often used for data analyses through clustering and visualisation. SOM’s mathematical model is inherently parallel. However, many implementations have not successfully exploited its parallelism because previous attempts often required cluster-like infrastructures. This article presents the parallel implementation of SOMs, particularly the batch map variant […]

August 18, 2011 by hgpu

This work describes a parallelizable optical flow field estimator based upon a modified batch version of the Self-Organizing Map (SOM). This estimator handles the ill-posedness in gradient-based motion estimation via a novel combination of regression and self-organization. The aperture problem is treated using an algebraic framework that partitions motion estimates obtained from regression into two […]

July 22, 2011 by hgpu