Analyzing the Performance of Multilayer Neural Networks for Object Recognition
This work addresses a knowledge gap for computer vision practitioners applying CNNs, but it is incremental as it builds on existing CNN successes without introducing a new method.
The paper tackles the problem of understanding the features learned by convolutional neural networks (CNNs) for object recognition, as they are less understood compared to engineered representations like SIFT and HOG, and it experimentally probes aspects of CNN feature learning to provide evidence-backed intuitions for practitioners.
In the last two years, convolutional neural networks (CNNs) have achieved an impressive suite of results on standard recognition datasets and tasks. CNN-based features seem poised to quickly replace engineered representations, such as SIFT and HOG. However, compared to SIFT and HOG, we understand much less about the nature of the features learned by large CNNs. In this paper, we experimentally probe several aspects of CNN feature learning in an attempt to help practitioners gain useful, evidence-backed intuitions about how to apply CNNs to computer vision problems.