An Iterative Pruning Approach of Neural Network for Proficient Noise Cancellation
Shashi Kant Dargar1, Himanshu Purohit2, S C Mahajan3
1Shashi Kant Dargar, Department of E & C Engineering, Sir Padampat Singhania University, India.
2Himanshu Purohit, Department of E & C Engineering, Sir Padampat Singhania University, India.
3Prof. S C Mahajan, Dean Academics, Sobasaria Engineering College, Sikar, India.
Manuscript received on March 12, 2013. | Revised Manuscript received on April 15, 2013. | Manuscript published on April 30, 2013. | PP: 7-9 | Volume-2, Issue-4, April 2013. | Retrieval Number: D1230042413/2013©BEIESP
Open Access | Ethics and Policies | Cite
© 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: For Active Noise Cancellations various algorithms run and result in standard output and better performance. Technique to pruning is related to prune nonfunctional neuron’s (algorithm) from ANC network, which makes only best neurons responsible for noise cancellation. Neurons are classified as different algorithms. Performance of neurons depends upon instantaneous surrounding conditions. So a proficient novel approach for ANC has been proposed. The wiener filter based on least means squared (LMS) algorithm family is most sought after solution of ANC. This family includes LMS, NLMS, VSLMS, VSNLMS, VFXLMS, FX- sLMS and many more. Some of these are nonlinear algorithm, which provides better solution for non linear noisy environment. The components of the ANC systems like microphones and loudspeaker exhibit nonlinearities themselves. The non linear transfer function create worse situation. For example, FX-sLMS algorithm behaves well than the second order VFXLMS algorithm in conditions of non-minimum phase and most important is the mean square error. The classical approach to RBF implementation is to fix the number of hidden neurons based on some property of the input data, and estimate the weights connecting the hidden and output neurons using linear least square method.
Keywords: Least Mean Square; RBF neural network; Artificial neural Network, Filter bank design, ANC.