Abstract

Recently, deep convolutional neural networks (CNN) with inception modules have attracted much attention due to their excellent performances on diverse domains. Nevertheless, the basic CNN can only capture a univariate feature, which is essentially linear. It leads to a weak ability in feature expression, further resulting in insufficient feature mining. In view of this issue, researchers incessantly deepened the network, bringing parameter redundancy and model over-fitting. Hence, whether we can employ this efficient deep neural network architecture to improve CNN and enhance the capacity of image recognition task still remains unknown. In this paper, we introduce spike-and-slab units to the modified inception module, enabling our model to capture dual latent variables and the average and covariance information. This operation further enhances the robustness of our model to variations of image intensity without increasing the model parameters. The results of several tasks demonstrated that dual variable operations can be well-integrated into inception modules, and excellent results have been achieved.

Highlights

  • The convolution neural network (CNN) significantly enhances the ability of image recognition by simulating the human brain

  • The research of image recognition based on convolutional neural networks (CNN) can significantly facilitate the development of artificial intelligence (AI)

  • We proposed an improved inception module to explore the strategy of constructing an optimal local sparse architecture of a CNN by utilizing the available dense modules

Read more

Summary

Introduction

The convolution neural network (CNN) significantly enhances the ability of image recognition by simulating the human brain. Traditional methods that mainly rely on artificial features result in a weak capacity to learn advanced feature information from the image. CNN owns a powerful expression and generalization ability in analyzing and interpreting data. How to robustly learn effective representations from complex images of different sizes still remains unsolved. The research of image recognition based on CNN can significantly facilitate the development of artificial intelligence (AI) technology. GoogLeNets [1], consisting of multiple inception modules, are typical and successful

Objectives
Methods
Results
Discussion
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call