ANALYSIS AND EXPERIMENTAL RESEARCH OF DEPENDENCE BETWEEN NEURAL NETWORKS TRAINING AND TRAINING PARAMETERS SELECTION
Abstract
The article offers an analysis of the results obtained from experimental research into the impact that the major parameters in training selection have on the general training capacity and recognition capacity in neural networks; there are also some recommendations offered regarding efficient development of large training selections containing graphic images of object categories.
About the Authors
Nikita Alekseyevich Lagunov
NCFU
Russian Federation
Russian Federation
Oksana Stanislavovna Mezentseva
NCFU
Russian Federation
Russian Federation
References
1. Ullman S., Vidal-Naquet M. Visual features of intermediate complexity and their use in classification // Nature Neuroscience, 2002. 490 c. URL: https://courses.csail.mit.edu/6.803/pdf/features.pdf.
2. Yann LeCun, Fu Jie Huang. Learning Methods for Generic Object Recognition with Invariance to Pose and Lighting // IEEE Computer Society Conference on Computer Vision and Pattern Recognition (volume 2), 2004. 104 с. URL: https://courses.csail.mit.edu/6.803/pdf/features.pdf.
3. Nicolas Pinto, David Cox, James DiCarlo. Why is Real-World Visual Object Recognition Hard? // {PLoS} Computational Biology, 2008. 27 c. URL: http://dicarlolab.mit.edu/sites/dicarlolab.mit.edu/files/pubs/Pinto%20et%20 al%202008.pdf.
4. James DiCarlo. How Does the Brain Solve Visual Object Recognition? Neuron. Cell-Press, 2012. 415 c.
5. Weber M., Welling M., Perona P. Unsupervised learning of models for recognition, 2000. URL: http://www. vision.caltech.edu/CNS179/papers/ Perona00.pdf.
Review
For citations:
Lagunov N.A., Mezentseva O.S. ANALYSIS AND EXPERIMENTAL RESEARCH OF DEPENDENCE BETWEEN NEURAL NETWORKS TRAINING AND TRAINING PARAMETERS SELECTION. Newsletter of North-Caucasus Federal University. 2014;(5):15-21. (In Russ.)
Views:
82
Email this article 