A Parallel Online Learning Algorithm for Anisotropic Digital Image Restoration

Bharti Patel; Rajesh Tiwari

doi:10.37591/rtpc.v2i1.314

Abstract

Previously, in the computational scenarios where the training data has already been made available for the learning algorithm, the kernel based algorithm such as support vector machines have made significant advances in the machine learning field. Though there has been little advancement made in learning scenario at online settings to facilitate the various real-time applications at the user end. In this study we discussed the issues with the online learning and propose a dynamic learning approach to facilitate the digital image restoration to overcome the online learning due to limited bandwidth and latency problem and thus consequently process the information in real time while on it is on route towards the receiver’s end. The algorithm distributes the batch jobs with respect to the cache miss rate and regressively simulates to adjust with the network latency. This association is effective in reproducing the lost or corrupted information over the network and help not only in retrieving the data but to process the jobs in a dynamic networked environment.Keywords: On-line learning and relative loss bounds, bandwidth, latency, multi-processing Cite this Article:Bharti Patel, Rajesh Tiwari. A Parallel Online Learning Algorithm for Anisotropic Digital Image Restoration. Recent Trends in Parallel Computing. 2015; 2(1): 21–26p.