Abstract:In the world of big data, extracting meaningful insights from large and continually growing distributed datasets is a major challenge. Classical clustering algorithms are effective at identifying clusters with convex structures. However, they fall short in identifying arbitrary-shaped clusters (more irregular and complex patterns), which are often encountered in real-world applications. The process of identifying non-convex cluster representations from very large and growing datasets is a challenge. It is further compounded by the distributed nature of the data, necessitating complex computations across multiple devices. Support Vector Clustering (SVC) is a much-celebrated algorithm capable of finding arbitrarily shaped clusters. However, the major limitation of this algorithm is that it will not scale to large volumes of data as the time and space complexity is high. The second limitation of the SVC algorithm is the requirement for large computation time in finding cluster structures. The adoption of a coreset based methodology is required for finding the true representation of the underlying large datasets. The implementation of hierarchical clustering on these distributed coresets, unlocks the potential to uncover a structured hierarchy of abstractions across the disseminated data. Moreover, a distance-based clustering approach guarantees the identification of clusters with diverse and arbitrary shapes, providing a robust framework for detecting complex structures. This research utilizes the Core Vector Machine (CVM) approach using an approximate Minimum Enclosing Ball (MEB) algorithm to efficiently address the complexities inherent in traditional SVC. Additionally, an enhanced medoid algorithm is employed for cluster head identification across the data sources. Hierarchical clustering is performed in the Reproducing Kernel Hilbert Space (RKHS) using cosine similarity distance matrices. This is used to identify compact non-convex clusters within distributed datasets. Performance assessment involves benchmarking our approach against state-of-the-art improved SVC algorithms using large datasets. The outcomes validate the superior performance of our approach compared to existing methods.

Distributed training of multiclass conic-segmentation support vector machines on communication constrained networks

A Fast-Convergence Distributed Support Vector Machine in Small-Scale Strongly Connected Networks

Communication and Energy-Constrained Neighbor Selection for Distributed Cooperative Localization

Distributed Online Semi-Supervised Support Vector Machine

Distributed Online One-Class Support Vector Machine for Anomaly Detection over Networks

High-Dimensional Distributed Sparse Classification with Scalable Communication-Efficient Global Updates

Parallelizing Support Vector Machines on Distributed Computers

Communication‐efficient distributed large‐scale sparse multinomial logistic regression

Parallel and Distributed Structured SVM Training

Communication-Efficient Distributed Learning via Sparse and Adaptive Stochastic Gradient

D-SVM over Networked Systems with Non-Ideal Linking Conditions

Dynamic Collaborative Support Vector Machine For Target Classification In Wireless Sensor Networks

GADGET SVM: A Gossip-bAseD sub-GradiEnT Solver for Linear SVMs

Scalable decision fusion algorithm for enabling decentralized computation in distributed, big data clustering problems

Chance constrained conic-segmentation support vector machine with uncertain data

Decentralized Edge Learning via Unreliable Device-to-Device Communications

Distributed ADMM with Synergetic Communication and Computation.

DCSVM: fast multi-class classification using support vector machines

Collaborative hybrid classifier learning with ant colony optimization in wireless multimedia sensor networks

RCD-SGD: Resource-Constrained Distributed SGD in Heterogeneous Environment via Submodular Partitioning

SDCC: Software-Defined Collective Communication for Distributed Training