Handwritten signatures recognition using Liquid State Machine
Souaad Belhia, Mohamed Benyettou, Ahmed Lehireche · International Journal of Biometrics · 2011
In this work, we checked the possibility of using Biological Neural Networks to analyse the complex multi-dimensional features for purposes of data recognition and classification. We investigated a recently proposed model ‘Liquid State Machine (LSM) using spiking neural network’ and its applicability for recognition of the ‘handwritten signature problem’. This project includes complex data analysis, adaptation of visual features to the neural microcircuit and comparison of the results with baseline algorithms. Experiments show that 90% correct classification rate can be achieved on a database of over 8000 signature images.