Parallel and accurate <i>k</i>‐means algorithm on CPU‐GPU architectures for spectral clustering

He, Guanlin; Vialle, Stéphane; Baboulin, Marc

doi:10.1002/cpe.6621

Cited by 6 publications

(4 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To meet the requirements for the simultaneous and accurate organization of very large-scale datasets, we have presented a three-step technique. By repeatedly running K-means with a customizable number of clusters, we use an uncommon strategy to select proxied samples in the first step [19]. To refine the selected cases based on their outlier scores, an outlier search method is also applied.…”

Section: Discussionmentioning

confidence: 99%

Large Scale Data Using K-Means

Belhaj¹,

zaib

Ourabah

2023

MJBD

View full text Add to dashboard Cite

Regular data base questioning tactics are insufficient to extract meaningful data due to the exponential expansion of high layered datasets; therefore, analysts nowadays are forced to build new processes to satisfy the increased needs. Because of the development in the number of data protests as well as the expansion in the number of elements/ascribes, such vast articulation data leads to numerous new computational triggers. To increase the effectiveness and accuracy of mining activities on highly layered data, the data should be preprocessed using a successful dimensionality decrease technique. So we have collected ideas of different researchers. In several fields, cluster analysis has recently gained popularity as a method for data analysis. A popular parceling-based clustering method called K-means searches for a certain number of clusters that may be found by their centroids. However, the results are quite dependent on the original cluster focus sites. Once more, the number of distance calculations significantly grows as the complexity of the data increases. This is because building a high-precision model frequently necessitates a sizable and dispersed preparatory set. A large preparation set could also need a significant amount of preparation time. There is a trade-off between speed and accuracy when creating orders, especially for large data sets. Vector data are frequently clustered, packed, and summed using the k-means approach. We provide No Concurrent Specific Clumped K-means, a rapid and memory-effective GPU-based approach for cautious k-means (ASB K-means). In contrast to previous GPU-based k-means methods, which require stacking the entire dataset onto the GPU for clustering, our methodology may be tailored to consume far less GPU RAM than the size of the complete dataset. As a result, we may cluster datasets that are bigger than the available RAM. In order to effectively handle large datasets, the method employs a clustered architecture and applies the triangle disparity in each k-means focus to eliminate a data point on the off chance that its enrollment task, or the cluster it is a member of, remains unchanged. As a result, fewer data guides have to be sent between the Slam of the computer processor and the global memory of the GPU.

show abstract

Section: Discussionmentioning

confidence: 99%

Large Scale Data Using K-Means

Belhaj¹,

zaib

Ourabah

2023

MJBD

View full text Add to dashboard Cite

show abstract

“…k ‐Means is a standard algorithm for clustering data used as the final step for high‐quality spectral clustering. To overcome the scalability challenge when processing large datasets, the authors of paper 2 propose to apply also the k ‐means algorithm as a preprocessing task to reduce the input data instances. Additionally, parallel optimization techniques are introduced to improve the efficiency of the k ‐means algorithm on CPU and GPU.…”

Section: Accepted Papers For the Special Issue: Summarymentioning

confidence: 99%

Algorithmic and software development advances for next‐generation heterogeneous platforms

Wyrzykowski

Ciorba

2022

Concurrency and Computation

View full text Add to dashboard Cite

Heterogeneity is emerging as one of the most profound and challenging characteristics of today's and tomorrow's parallel and distributed computing environments, presenting new and exciting opportunities for their development. Most modern computing systems are heterogeneous, either for organic reasons because components grew independently, as is the case of desktop grids, by design to leverage the strength of specific hardware, as is the case of accelerated systems, or both. The impact of heterogeneity on all forms of parallel and distributed computing is increasing rapidly.Traditional algorithms, programming environments, and tools designed for legacy homogeneous systems will at best achieve a small fraction of the efficiency and the potential performance expected from parallel computing in tomorrow's highly diversified and mixed architectures. Innovative ideas, fresh models, novel algorithms, and other specialized or unified programming environments and tools are needed to efficiently use these new and increasingly diverse computing systems-for accelerating scientific discovery and impactful innovation.The International Workshop on Algorithms, Models and Tools for Parallel Computing on Heterogeneous Platforms (HeteroPar) has been the premier forum over the last 20 years, bringing together researchers to discuss these challenges and the solutions. The wide range of topics includes achieving performance portability on heterogeneous architectures, advances in software environments that facilitate efficient use of heterogeneous systems, performance and energy optimization of numerical and machine learning algorithms on heterogeneous platforms, to name a few.The works presented at the HeteroPar'2020 workshop covered topics clearly exhibiting the significance and growth of the heterogeneous computing field. However, one general trend is apparent: the broad adoption of Graphics Processing Units (GPU) accelerators. Over the last decade, GPUs have been established as the main powerhouse in leadership supercomputers and an invaluable component to accelerate computations for a vast spectrum of applications-from numerical linear algebra libraries powering computational science to various machine learning workloads. This trend is evidenced by the increasing number of GPU-related publications submitted to HeterPar and supported by growing diversity within the GPU world, where AMD accelerator architectures start to compete with Nvidia's comprehensive solutions, along with the third GPU accelerator option-from Intel-available soon.This special issue of Concurrency and Computation: Practice and Experience contains six selected papers from the HeteroPar'2020 workshop.We hope you find them interesting and stimulating new ideas and future advancements for next-generation heterogeneous platforms.

show abstract

“…Although many papers have been published on accelerating k-means using GPUs, almost all of them are based on the standard k-means algorithm by applying different optimization techniques on various steps and require the whole dataset to be loaded into the GPU's global memory (Kruliš and Kratochvíl 2020;He et al 2022;Taylor and Gowanlock 2021). To the best of our knowledge, massively parallel processing of the k-means clustering algorithm accelerated with the triangle inequality has not been reported in the literature, especially an algorithm that is capable of handling datasets larger than the global memory of the GPU.…”

Section: Gpu-based K-means Implementationsmentioning

confidence: 99%

Large scale K-means clustering using GPUs

Frank

Pfahringer

2022

Data Min Knowl Disc

View full text Add to dashboard Cite

The k-means algorithm is widely used for clustering, compressing, and summarizing vector data. We present a fast and memory-efficient GPU-based algorithm for exact k-means, Asynchronous Selective Batched K-means (ASB K-means). Unlike most GPU-based k-means algorithms that require loading the whole dataset onto the GPU for clustering, the amount of GPU memory required to run our algorithm can be chosen to be much smaller than the size of the whole dataset. Thus, our algorithm can cluster datasets whose size exceeds the available GPU memory. The algorithm works in a batched fashion and applies the triangle inequality in each k-means iteration to omit a data point if its membership assignment, i.e., the cluster it belongs to, remains unchanged, thus significantly reducing the number of data points that need to be transferred between the CPU’s RAM and the GPU’s global memory and enabling the algorithm to very efficiently process large datasets. Our algorithm can be substantially faster than a GPU-based implementation of standard k-means even in situations when application of the standard algorithm is feasible because the whole dataset fits into GPU memory. Experiments show that ASB K-means can run up to 15x times faster than a standard GPU-based implementation of k-means, and it also outperforms the GPU-based k-means implementation in NVIDIA’s open-source RAPIDS machine learning library on all the datasets used in our experiments.

show abstract

Parallel and accurate k‐means algorithm on CPU‐GPU architectures for spectral clustering

Cited by 6 publications

References 19 publications

Large Scale Data Using K-Means

Large Scale Data Using K-Means

Algorithmic and software development advances for next‐generation heterogeneous platforms

Large scale K-means clustering using GPUs

Contact Info

Product

Resources

About