Open Access Open Access  Restricted Access Subscription Access

Deep Convolutional Auto-Multiplexers for Emotion Recognition from Facial Expressions

T MUNI REDDY, R. P. Singh

Abstract


Now a day’s one of the unsolved problem in computer vision is recognizing or understanding other people's emotions and feelings. Deep Convolutional Neural Networks (CNN) has tried to be economical in feeling recognition issues. The good degree of performance achieved by these classifiers can be attributed to their ability to self-learn a down-sampled feature vector that retains abstraction info through filter kernels in convolutional layers. In this paper we have a tendency to explore the impact of coaching the initial weights in associate unsupervised manner. we have a tendency to study the result of pre-training a Deep CNN as a Convolutional Auto-Multiplexer (CAM) in a very greedy layer-wise unsupervised fashion for emotion recognition mistreatment facial features pictures. Once trained with at random initialized weights, our CNN feeling recognition model achieves a performance rate of 92.16% on the Karolinska Directed Emotional Faces (KDEF) dataset. In distinction, by using this pre-trained, the performance will increase to 93.52%. Pre-training our CNN as a CAM conjointly reduces coaching time marginally.

 

Keywords: Emotion, face expression, convolutional auto-multiplex (CAM), MAM pooling, DCNN


Full Text:

PDF

References


Lewis M., Haviland-Jones J.M., Barrett L.F. Handbook of Emotions. Contemporary Sociology, vol. 24 no.3, 298p, may 1995.

ChavhanA., Chavan S., Dahe S., Chibhade S. A Neural Network Approach for Real Time Emotion Recognition. Ijarcce, vol. 4, no. 3, 259–263p, 2015.

LundqvistD., Flykt A., O¨ hman A. The Karolinska Directed Emotional Faces. KDEF CD ROM from Department of Clinical Neuroscience, Psycology section. Karolinska Institutet, 3–5p, 1998.

Bengio Y. Learning Deep Architectures for AI. Foundations and Trends R _ in Machine Learning, vol. 2, no. 1, 1–127p, 2009.

Kr¨ahenb¨uhl P., Doersch C., Donahue J., Darrell T. Data-dependent Initializations of Convolutional Neural Networks. International Conference on Computer Vision, 1–12p, nov 2015

Romero A., Ballas N., Kahou S.E., et al. FitNets: Hints for Thin Deep Nets. 1–13p, dec 2014.

Srivastava R.K., Greff K., Schmidhuber J. Highway Networks. arXiv:1505.00387 [cs], may 2015.

Mishkin D., Matas J. All you need is a good init. Computers & Mathematics with Applications, vol. 31, no. 11, p. 135, nov 2015.

He K., Zhang X., Ren S., Sun J. Delving Deep into Rectifiers: Surpassing Human-Level Performance on Image Net Classification. feb 2015.

Maas A.L., Hannun A.Y., Ng A.Y. Rectifier Nonlinearities Improve Neural Network Acoustic Models. Proceedings of the 30th International Conference on Machine Learning, vol. 28, 6p, 2013.

Xu B., Wang N., Chen T., Li M. Empirical Evaluation of Rectified Activations in Convolutional Network. ICML Deep Learning Workshop. 1–5p, may 2015.

Reddy T.M., Sing R.P. Real Time Human Emotion Recognition Using Artificial Neural Networks. IJARCSSE, Vol.8, Issue 4, April 2018.

Ioffe S., Szegedy C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. feb 2015.

Erhan D., Bengio Y., Courville A., Mazagol P.A., Vincent P. Representation Learning: A Review and New Perspectives. Journal of Machine Learning Research, vol. 11, 625–660p, June 2010.

Hinton G.E., Osindero S., Teh Y.W. A fast learning algorithm for deep belief nets. Neural computation, vol. 18, no. 7, 1527–54p, July 2006.

Abdel-Hamid O., Deng L., Yu D. Exploring convolutional neural network structures and optimization techniques for speech recognition. INTERSPEECH, 2013.

MasciJ., Meier U., Cirean D., Schmidhuber J. Stacked Convolutional Auto-Encoders for Hierarchical Feature Extraction. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2011, vol. 6791 LNCS, no. PART 1, 52–59p.

BurkertP., Trier F., Afzal, A. Dengel M.Z., Liwicki M. DeXpression: Deep convolutional neural network for expression recognition. arXiv preprint, 1–8p, 2015.


Refbacks

  • There are currently no refbacks.