A comparative study between standard Back Propagation and Resilient Propagation on snake identification accuracy

Abstract

Identifying types of snakes is crucial for appropriate anti-venom administration. We developed a Snake Bite Diagnosing System based on standard Back Propagation and Resilient Propagation Neural Networks. These systems were capable in differentiating between venomous and non-venomous snakes. The accuracy of both systems were analyzed and compared. The post development comparative studies revealed that the Resilient Propagation technique yielded 83.33%, 90.00% and 90.00% respectively for mean squared error, number of epoch and parameter setting. Whereas the Standard Back Propagation produced 85.00%, 88.30% and 86.87% respectively. High accuracy in both systems enables early identification of type of snake and immediate specific anti-venom can be administered. Hence, reduces the rate of morbidity and mortality.