Region Based Image Preprocessor for Feed-Forward Perceptron Based Systems
In this paper, we investigate the notion that there may be alternate methods, beyond typical rectilinear interpolations such as Bilinear Interpolation, that have a greater suitability for use in visual/image preprocessors for Artificial Neural Networks. We present a novel method for down-sampling image data in preparation for a Feed-Forward Perceptron system assisted by a neural usefulness metric, inspired by those common to pruning algorithms. This new method achieves greater accuracy compared to the same system using by Bilinear Interpolation, and has a reduced computational time.
KeywordsArtificial neural networks Preprocessor Image processing Salience Relevance assessment
Unable to display preview. Download preview PDF.
- 1.Shibata, K., Utsunomiya, H.: Discovery of pattern meaning from delayed rewards by reinforcement learning with a recurrent neural network. In: The 2011 International Joint Conference on Neural Networks (IJCNN), pp. 1445–1452 (2011)Google Scholar
- 2.Davies, S., Patterson, C., Galluppi, F., Rast, A., Lester, D., Furber, S.: Interfacing real-time spiking i/o with the spinnaker neuromimetic architecture. Australian Journal of Intelligent Information Processing System 11, 7–11 (2010)Google Scholar
- 4.Mozer, M.C., Smolensky, P.: Skeletonization: a technique for trimming the fat from a network via relevance assessment. Advances in Neural Information Processing Systems 1, 107–115 (1989)Google Scholar