| Abstrakt: |
Convolutional Neural Networks (CNN) is a deep learning algorithm which is used with image-related data and it is used for recognizing and analysing their features. Traditionally, the Backpropagation (BP) algorithm is used to train them. But one of the key limitations of this algorithm is its slow learning rate which in turn increases the training and computational time of the CNN and as data size gets larger, the training time will keep on increasing. Here, by using the Extreme Learning Machine (ELM) algorithm, this limitation of BP algorithm can be countered, the training time of the model will become less and the computation of results will be done much faster. To prove our point, we have done a comparative study where our work attempts to address different learning algorithms for the feed-forward neural networks. Here, we compare the performance of Convolutional Neural Networks when it is pipelined with three algorithms i.e., BP, Support Vector Machine (SVM) and ELM. After comparing the performance of these three learning algorithms with proper results, by using ELM, the time taken to deliver the results is much faster compared to the other algorithms used for comparison in our study. |
Full text from SpringerLink