3 lines
710 B
TeX
3 lines
710 B
TeX
\chapter*{Abstract}
|
|
Deep learning integrates feature extraction directly into the traditional neural network architecture, improving overall performance. This paper explores the benefits of deep learning by using the MNIST handwritten digit dataset. We compare two different network configurations: one with feature extraction and feed-forward classification, and the other with only the feed-forward classification. Our results demonstrate that as the size of the feed-forward network is reduced, its classification performance decreases. However, by leveraging feature extraction, we are able to retain classification power, showing the value of deep learning in improving performance with smaller networks.
|