Graphics Processing Unit Parallel Implementation Research Articles

GPU Parallel Implementation of Spatially Adaptive Hyperspectral Image Classification

Image classification is a very important tool for remotely sensed hyperspectral image processing. Techniques able to exploit the rich spectral information contained in the data, as well as its spatial-contextual information, have shown success in recent years. Due to the high dimensionality of hyperspectral data, spectral-spatial classification techniques are quite demanding from a computational viewpoint. In this paper, we present a computationally efficient parallel implementation for a spectral-spatial classification method based on spatially adaptive Markov random fields (MRFs). The method learns the spectral information from a sparse multinomial logistic regression classifier, and the spatial information is characterized by modeling the potential function associated with a weighted MRF as a spatially adaptive vector total variation function. The parallel implementation has been carried out using commodity graphics processing units (GPUs) and the NVIDIA's Compute Unified Device Architecture. It optimizes the work allocation and input/output transfers between the central processing unit and the GPU, taking full advantages of the computational power of GPUs as well as the high bandwidth and low latency of shared memory. As a result, the algorithm exploits the massively parallel nature of GPUs to achieve significant acceleration factors (higher than 70x) with regards to the serial and multicore versions of the same classifier on an NVIDIA Tesla K20C platform.

GPU Parallel Implementation of Isometric Mapping for Hyperspectral Classification

Manifold learning algorithms such as the isometric mapping (ISOMAP) algorithm have been widely used in the analysis of hyperspectral images (HSIs), for both visualization and dimension reduction. As advanced versions of the traditional linear projection techniques, the manifold learning algorithms find the low-dimensional feature representation by nonlinear mapping, which can better preserve the local structure of the original data and thus benefit the data analysis. However, the high computational complexity of the manifold learning algorithms hinders their application in HSI processing. Although there are a few parallel implementations of manifold learning approaches that are available in the remote sensing community, they have not been designed to accelerate the eigen-decomposition process, which is actually the most time-consuming part of the manifold learning algorithms. In this letter, as a case study, we discuss the graphics processing unit parallel implementation of the ISOMAP algorithm. In particular, we focus on the eigen-decomposition process and verify the applicability of the proposed method by validating the embedding vectors and the subsequent classification accuracies. The experimental results obtained on different HSI data sets show an excellent speedup performance and consistent classification accuracy compared with the serial implementation.

Parallel Implementation of the Recursive Least Square for Hyperspectral Image Compression on GPUs

Compression is a very important technique for remotely sensed hyperspectral images. The lossless compression based on the recursive least square (RLS), which eliminates hyperspectral images` redundancy using both spatial and spectral correlations, is an extremely powerful tool for this purpose, but the relatively high computational complexity limits its application to time-critical scenarios. In order to improve the computational efficiency of the algorithm, we optimize its serial version and develop a new parallel implementation on graphics processing units (GPUs). Namely, an optimized recursive least square based on optimal number of prediction bands is introduced firstly. Then we use this approach as a case study to illustrate the advantages and potential challenges of applying GPU parallel optimization principles to the considered problem. The proposed parallel method properly exploits the low-level architecture of GPUs and has been carried out using the compute unified device architecture (CUDA). The GPU parallel implementation is compared with the serial implementation on CPU. Experimental results indicate remarkable acceleration factors and real-time performance, while retaining exactly the same bit rate with regard to the serial version of the compressor.

GPU parallel implementation for asset-liability management in insurance companies

In this work we present a stochastic asset liability management (ALM) model for a life insurance company together with its numerical simulation, based in a Monte Carlo balance sheet projection, and we carry out its efficient parallel computation using graphics processing units (GPUs) hardware. The liabilities of the company consist of a portfolio comprising with-profit life insurance policies, that evolve according to the policyholder saving account, surrender and biometric models. On the asset side, we mainly consider bonds, equity and cash, so that appropriate stochastic models are considered for their evolution. We consider some innovations with respect to literature in the modeling of the surrenders of the policyholders. Another important innovative aspect comes from the implementation of ALM in the new high performance computing architectures provided by GPUs technology. Numerical results illustrate the high speed up of the calculus by using GPUs and the coherence of the computations (asset evolution, default probabilities and so on).

Parallel computation for blood cell classification in medical hyperspectral imagery

With the advantage of fine spectral resolution, hyperspectral imagery provides great potential for cell classification. This paper provides a promising classification system including the following three stages: (1) band selection for a subset of spectral bands with distinctive and informative features, (2) spectral-spatial feature extraction, such as local binary patterns (LBP), and (3) followed by an effective classifier. Moreover, these three steps are further implemented on graphics processing units (GPU) respectively, which makes the system real-time and more practical. The GPU parallel implementation is compared with the serial implementation on central processing units (CPU). Experimental results based on real medical hyperspectral data demonstrate that the proposed system is able to offer high accuracy and fast speed, which are appealing for cell classification in medical hyperspectral imagery.

Parallel Implementation of Sparse Representation Classifiers for Hyperspectral Imagery on GPUs

Classification is one of the most important analysis techniques for hyperspectral image analysis. Sparse representation is an extremely powerful tool for this purpose, but the high computational complexity of sparse representation-based classification techniques limits their application in time-critical scenarios. To improve the efficiency and performance of sparse representation classification techniques for hyperspectral image analysis, this paper develops a new parallel implementation on graphics processing units (GPUs). First, an optimized sparse representation model based on spatial correlation regularization and a spectral fidelity term is introduced. Then, we use this approach as a case study to illustrate the advantages and potential challenges of applying GPU parallel optimization principles to the considered problem. The first GPU optimization algorithm for sparse representation classification (SRCSC_P) of hyperspectral images is proposed in this paper, and a parallel implementation of the proposed method is developed using compute unified device architecture (CUDA) on GPUs. The GPU parallel implementation is compared with the serial and multicore implementations on CPUs. Experimental results based on real hyperspectral datasets show that the average speedup of SRCSC_P is more than $\mathbf{130} \times$ , and the proposed approach is able to provide results accurately and fast, which is appealing for computationally efficient hyperspectral data processing.

TuCCompi: A Multi-layer Model for Distributed Heterogeneous Computing with Tuning Capabilities

During the last decade, parallel processing architectures have become a powerful tool to deal with massively-parallel problems that require high performance computing (HPC). The last trend of HPC is the use of heterogeneous environments, that combine different computational processing devices, such as CPU-cores and graphics processing units (GPUs). Maximizing the performance of any GPU parallel implementation of an algorithm requires an in-depth knowledge about the GPU underlying architecture, becoming a tedious manual effort only suited for experienced programmers. In this paper, we present TuCCompi, a multi-layer abstract model that simplifies the programming on heterogeneous systems including hardware accelerators, by hiding the details of synchronization, deployment, and tuning. TuCCompi chooses optimal values for their configuration parameters using a kernel characterization provided by the programmer. This model is very useful to tackle problems characterized by independent, high computational-load independent tasks, such as embarrassingly-parallel problems. We have evaluated TuCCompi in different, real-world, heterogeneous environments using the all-pair shortest-path problem as a case study.

Real-time implementation of optimized maximum noise fraction transform for feature extraction of hyperspectral images

We present a parallel implementation of the optimized maximum noise fraction (G-OMNF) transform algorithm for feature extraction of hyperspectral images on commodity graphics processing units (GPUs). The proposed approach explored the algorithm data-level concurrency and optimized the computing flow. We first defined a three-dimensional grid, in which each thread calculates a sub-block data to easily facilitate the spatial and spectral neighborhood data searches in noise estimation, which is one of the most important steps involved in OMNF. Then, we optimized the processing flow and computed the noise covariance matrix before computing the image covariance matrix to reduce the original hyperspectral image data transmission. These optimization strategies can greatly improve the computing efficiency and can be applied to other feature extraction algorithms. The proposed parallel feature extraction algorithm was implemented on an Nvidia Tesla GPU using the compute unified device architecture and basic linear algebra subroutines library. Through the experiments on several real hyperspectral images, our GPU parallel implementation provides a significant speedup of the algorithm compared with the CPU implementation, especially for highly data parallelizable and arithmetically intensive algorithm parts, such as noise estimation. In order to further evaluate the effectiveness of G-OMNF, we used two different applications: spectral unmixing and classification for evaluation. Considering the sensor scanning rate and the data acquisition time, the proposed parallel implementation met the on-board real-time feature extraction.

AN EVALUATION OF MULTIPLE FEED-FORWARD NETWORKS ON GPUs

The Graphics Processing Unit (GPU) originally designed for rendering graphics and which is difficult to program for other tasks, has since evolved into a device suitable for general-purpose computations. As a result graphics hardware has become progressively more attractive yielding unprecedented performance at a relatively low cost. Thus, it is the ideal candidate to accelerate a wide variety of data parallel tasks in many fields such as in Machine Learning (ML). As problems become more and more demanding, parallel implementations of learning algorithms are crucial for a useful application. In particular, the implementation of Neural Networks (NNs) in GPUs can significantly reduce the long training times during the learning process. In this paper we present a GPU parallel implementation of the Back-Propagation (BP) and Multiple Back-Propagation (MBP) algorithms, and describe the GPU kernels needed for this task. The results obtained on well-known benchmarks show faster training times and improved performances as compared to the implementation in traditional hardware, due to maximized floating-point throughput and memory bandwidth. Moreover, a preliminary GPU based Autonomous Training System (ATS) is developed which aims at automatically finding high-quality NNs-based solutions for a given problem.

Parallel GPU Implementation of Iterative PCA Algorithms

Principal component analysis (PCA) is a key statistical technique for multivariate data analysis. For large data sets, the common approach to PCA computation is based on the standard NIPALS-PCA algorithm, which unfortunately suffers from loss of orthogonality, and therefore its applicability is usually limited to the estimation of the first few components. Here we present an algorithm based on Gram-Schmidt orthogonalization (called GS-PCA), which eliminates this shortcoming of NIPALS-PCA. Also, we discuss the GPU (Graphics Processing Unit) parallel implementation of both NIPALS-PCA and GS-PCA algorithms. The numerical results show that the GPU parallel optimized versions, based on CUBLAS (NVIDIA), are substantially faster (up to 12 times) than the CPU optimized versions based on CBLAS (GNU Scientific Library).