Implementing a foveal-pit inspired filter in a Spiking Convolutional Neural Network: a preliminary study
| Autor: | Shriya T. P. Gupta, Basabdatta Sen Bhattacharya |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Difference of Gaussians Computer science Computer Science - Artificial Intelligence Computer Vision and Pattern Recognition (cs.CV) Computer Science - Computer Vision and Pattern Recognition 02 engineering and technology Convolutional neural network Reduction (complexity) 03 medical and health sciences 0302 clinical medicine Foveal Encoding (memory) 0202 electrical engineering electronic engineering information engineering Redundancy (engineering) I.2.10 business.industry I.4.5 Pattern recognition Filter (signal processing) Backpropagation I.4.10 Artificial Intelligence (cs.AI) 020201 artificial intelligence & image processing Artificial intelligence business 030217 neurology & neurosurgery |
| DOI: | 10.48550/arxiv.2105.14326 |
| Popis: | We have presented a Spiking Convolutional Neural Network (SCNN) that incorporates retinal foveal-pit inspired Difference of Gaussian filters and rank-order encoding. The model is trained using a variant of the backpropagation algorithm adapted to work with spiking neurons, as implemented in the Nengo library. We have evaluated the performance of our model on two publicly available datasets - one for digit recognition task, and the other for vehicle recognition task. The network has achieved up to 90% accuracy, where loss is calculated using the cross-entropy function. This is an improvement over around 57% accuracy obtained with the alternate approach of performing the classification without any kind of neural filtering. Overall, our proof-of-concept study indicates that introducing biologically plausible filtering in existing SCNN architecture will work well with noisy input images such as those in our vehicle recognition task. Based on our results, we plan to enhance our SCNN by integrating lateral inhibition-based redundancy reduction prior to rank-ordering, which will further improve the classification accuracy by the network. Comment: 8 pages, 8 figures, 4 tables. 2020 International Joint Conference on Neural Networks (IJCNN) |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|