Large-scale applications such as pattern recognition (PR) and especially based on deep learning techniques have been moving from centralized to decentralized or distributed platform to improve their scalability. Therefore, in the last few decades, researchers have put an enormous effort to implement different and robust deep learning techniques. Nevertheless, these techniques are focused for small and medium quantities of documents. In this paper, we propose the hybrid peer-to-peer (P2P), Grid computing and agent technology, that allows PR computations to scale out to multiple peers and grids and tolerate several types of faults. We performed an extensive experimental evaluation in the P2P–Grid–Agent Distributed Platform with a real large-scale dataset from the IFN/ENIT (Institute of Communications Technology (IFN)), Technical University Braunschweig, Germany, Ecole Nationale d’Ingénieurs de Tunis (ENIT), Tunisia. Results prove that our solution significantly reduces execution time when compared to traditional methods that achieve the same level of resilience and guarantee the performances of large-scale applications.
CITATION STYLE
Hassen, H., Zied, T., & Maher, K. (2019). The P2P–grid–agent distributed platform: A distributed and dynamic platform for developing and executing large-scale application based on deep learning techniques. In Smart Innovation, Systems and Technologies (Vol. 143, pp. 25–35). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-13-8303-8_3
Mendeley helps you to discover research relevant for your work.