Optimized Variational Bayesian Extreme Learning Machine Algorithm for Multimodal Biometric Recognition
Sandhya Tarar1, Vyomika Singh2, Vibhash Yadav3, Shekhar Singh4, Hemant Gupta5
1Sandhya Tarar, Department of Information and Communication Technology, Gautam Buddha University, Noida (Uttar Pradesh), India.
2Vyomika Singh, Department of Information and Communication Technology, Gautam Buddha University, Noida (Uttar Pradesh), India.
3Vibhash Yadav, Department of Information Technology, Rajkiya Engineering College, Banda (Uttar Pradesh), India.
4Shekhar Singh, Department of Computer Science and Engineering, Shri Venketeshwara University, Gajraula (Uttar Pradesh), India.
5Hemant Gupta, Department of Computer Science and Engineering, Carleton University, Canada North America.
Manuscript received on 22 March 2019 | Revised Manuscript received on 05 April 2019 | Manuscript Published on 11 April 2019 | PP: 36-43 | Volume-8 Issue-4C, April 2019 | Retrieval Number: D24180484C19/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In the thriving field of secure biometric systems, numerous advancements have been created and the need of the hour is Variational Bayesian Extreme Learning Machine (VBELM) which has an advantage in terms of time efficiency, speed, security and accuracy over traditional Extreme Learning Machine method (ELM). After observing the experimental results of Variational Bayesian Extreme Learning Machine (VBELM) we observe that testing accuracy, over fitting problem and recognition models are the issues and in order to address them, we curate an Optimized VBELM (OVBELM) which has opened doors for an exceptional performance in terms of improved statistical testing accuracy, improved recognition rates, execution time, reduced error rates and improved average fusion time.In this paper, optimized Variational Bayesian Extreme Learning Machine (VBELM) is based on local feature fusion of three modalities- Face, Fingerprint and Iris where appending iris as a third modality makes the system robust and secure. The optimized biometric recognition system which is trained on an artificial neural network (ANN) exhibits exceptional results after applying on 240 face images (40 people with 6 images for each individual) from FERET Face database (Facial Recognition Database), 240 fingerprint images( 40 people with 6 images for each individual) from FVC2002 fingerprint database and 240 iris images (40 people with 6 images for each individual) from UBIRIS database and result analysis depicts that the optimized VBELM (OVBELM) is having an edge over VBELM and traditional ELM duly reflected with improved execution time , testing accuracy, average fusion time and reduced error rates.
Keywords: Artificial Neural Network (ANN),Extreme Learning Machine (ELM), Feature based Fusion, Multimodal Biometrics System, Optimizedm Varia-tional Bayesian Extreme Learning Machine (OVBELM);Variational Bayesian Extreme Learning Machine (VBELM).
Scope of the Article: Machine Learning