Deep Learning through Convolutional Neural Networks for Classification of Image: A Novel Approach Using Hyper Filter
DOI:
https://doi.org/10.26438/ijcse/v7i6.164168Keywords:
Deep learning, Convolutional Neural Network, Image Classification, CIFAR-100, CIFAR-10Abstract
The convolutional neural networks (CNN) are artificial neural networks (ANN) having many similarities like layered architecture, neurons, activation function, and learning rate are some of them. There are some differences also like in CNN we can also deal with tensors which is the most distinguishing feature of CNN and these are just multidimensional 2D or 3D arrays. Another difference is layers in CNN are not same as in ANN. The common layers present in CNN are called as convolutional, relu and maxpool and these are generally connected sequentially so that the output of one layer acts as input to another layer. In the current article, the hybrid approach of filters or kernel is proposed and is giving better results in comparison to other kernel initializers like variance scaling normally used in CNN. The dataset used is CIFAR-100.
References
[1] Krizhevsky, A., Hinton, G., “Learning multiple layers of features from tiny images”, (2009).
[2] D. H. Hubel and T. N. Wiesel, “Receptive fields of single neurons in the cat’s striate cortex”, J. Physiol., vol. 148, no. 1, pp. 574–591, 1959.
[3] K. Fukushima, “Neocognitron: A self-organizing neural network model for a mechanism of pattern recog-nition unaffected by shift in position”, Biol. Cybern., vol. 36, no. 4, pp. 193–202, April, 1980.
[4] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recogni-tion,” Proc. IEEE, vol. 86, no. 11, pp. 2278–2324, Nov. 1998.
[5] C. B. Bishop, "Neural Networks for Pattern Recogni-tion", Oxford University Press, Oxford, 1995.
[6] Satish Kumar, "Neural Networks A Classroom Ap-proach", Tata McGraw Hill, 2013.
[7] https://keras.io
[8] D. E. Rumelhart, G.E. Hinton, R.J. Williams, "Learn-ing internal representation by error propagation", Paral-lel distributed processing: Explorations in the microstructure of cognition, Vol.1, Bradford books, Cambridge, MA, 1986.
[9] N. Rezazadeh, "Initialization of weights in deep belief neural network based on standard deviation of feature values in training data vectors", International Journal of Scientific Research in Computer Science and Engineering, Vol.5, Issue.4, pp.1-8, 2017
[10] Kshitij Tripathi, Rajendra G. Vyas, Anil K. Gupta, “The Classification of Data: A Novel Artificial Neural Network (ANN) Approach through Exhaustive Validation and Weight Initialization”, International Journal of Computer Sciences and Engineering, Vol.6, Issue.5, pp.241-254, 2018.
[11] Mohammad Jafari, Neda Abdollahi, Ali Amiri, Mahmood Fathy, "Generalization of Determinant Ker-nels for Non-Square Matrix and its Application in Video Retrieval", International Journal of Scientific Research in Computer Science and Engineering, Vol.3, Issue.4, pp.1-6, 2015
[12] Yu Liu,Yanming Guo, Theodoros Georgiou, Mi-chael S. Lew, “Fusion that matters: convolutional fusion networks for visual recognition”, Multimedia Tools Appl, 2018.
[13] Yanming Guo, Yu Liu, Erwin M. Bakker
Yuanhao Guo, Michael S. Lew, “CNN-RNN: a large-scale hierarchical image classification framework”, Multimed Tools Appl, 2018.
[14] Yi Zhou, Yue Bai, Shuvra S. Bhattacharyya and Heikki Huttunen," Elastic Neural Networks for Classifi-cation", arXiv, 2019
[15] Yanping Huang, Youlong Cheng, Dehao Chen, HyoukJoong Lee, Jiquan Ngiam, Quoc V. Le and Zhif-eng Chen,"GPipe: Efficient Training of Giant Neural Networks using Pipeline Parallelism", arXiv, 2018.
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors contributing to this journal agree to publish their articles under the Creative Commons Attribution 4.0 International License, allowing third parties to share their work (copy, distribute, transmit) and to adapt it, under the condition that the authors are given credit and that in the event of reuse or distribution, the terms of this license are made clear.
