Gradient control plays an important role in feed-forward networks applied to various computer vision tasks. Previous work has shown that Recurrent Highway Networks minimize the problem of vanishing or exploding gradients. They achieve this by setting the eigenvalues of the temporal Jacobian to 1 across the time steps. In this work, batch normalized recurrent highway networks are proposed to control the gradient flow in an improved way for network convergence. Specifically, the introduced model can be formed by batch normalizing the inputs at each recurrence loop. The proposed model is tested on an image captioning task using MSCOCO dataset. Experimental results indicate that the batch normalized recurrent highway networks converge faster and performs better compared with the traditional LSTM and RHN based models.
Paper Details
- Authors:
- Submitted On:
- 19 September 2017 - 1:59am
- Short Link:
- Type:
- Presentation Slides
- Event:
- Presenter's Name:
- Shagan Sah
- Paper Code:
- 3118
- Document Year:
- 2017
- Cite
Keywords
Subscribe
url = {http://sigport.org/2233},
author = {Chi Zhang; Thang Nguyen; Shagan Sah; Raymond Ptucha; Alexander Loui; Carl Salvaggio },
publisher = {IEEE SigPort},
title = {Presentation Slides for paper #3118},
year = {2017} }
T1 - Presentation Slides for paper #3118
AU - Chi Zhang; Thang Nguyen; Shagan Sah; Raymond Ptucha; Alexander Loui; Carl Salvaggio
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/2233
ER -
