Input Fast-Forwarding for Better Deep Learning. Ibrahim, A., Abbott, A. L., & Hussein, M. E. In Karray, F., Campilho, A., & Cheriet, F., editors, Image Analysis and Recognition, of Lecture Notes in Computer Science, pages 363–370, 2017. Springer International Publishing.
abstract   bibtex   
This paper introduces a new architectural framework, known as input fast-forwarding, that can enhance the performance of deep networks. The main idea is to incorporate a parallel path that sends representations of input values forward to deeper network layers. This scheme is substantially different from “deep supervision”, in which the loss layer is re-introduced to earlier layers. The parallel path provided by fast-forwarding enhances the training process in two ways. First, it enables the individual layers to combine higher-level information (from the standard processing path) with lower-level information (from the fast-forward path). Second, this new architecture reduces the problem of vanishing gradients substantially because the fast-forwarding path provides a shorter route for gradient backpropagation. In order to evaluate the utility of the proposed technique, a Fast-Forward Network (FFNet), with 20 convolutional layers along with parallel fast-forward paths, has been created and tested. The paper presents empirical results that demonstrate improved learning capacity of FFNet due to fast-forwarding, as compared to GoogLeNet (with deep supervision) and CaffeNet, which are 4×4×4\\times \ and 18×18×18\\times \ larger in size, respectively. All of the source code and deep learning models described in this paper will be made available to the entire research community (https://github.com/aicentral/FFNet).
@inproceedings{ibrahim_input_2017,
	title = {Input Fast-Forwarding for Better Deep Learning},
	rights = {All rights reserved},
	isbn = {978-3-319-59876-5},
	series = {Lecture Notes in Computer Science},
	abstract = {This paper introduces a new architectural framework, known as input fast-forwarding, that can enhance the performance of deep networks. The main idea is to incorporate a parallel path that sends representations of input values forward to deeper network layers. This scheme is substantially different from “deep supervision”, in which the loss layer is re-introduced to earlier layers. The parallel path provided by fast-forwarding enhances the training process in two ways. First, it enables the individual layers to combine higher-level information (from the standard processing path) with lower-level information (from the fast-forward path). Second, this new architecture reduces the problem of vanishing gradients substantially because the fast-forwarding path provides a shorter route for gradient backpropagation. In order to evaluate the utility of the proposed technique, a Fast-Forward Network ({FFNet}), with 20 convolutional layers along with parallel fast-forward paths, has been created and tested. The paper presents empirical results that demonstrate improved learning capacity of {FFNet} due to fast-forwarding, as compared to {GoogLeNet} (with deep supervision) and {CaffeNet}, which are 4×4×4\{{\textbackslash}times \} and 18×18×18\{{\textbackslash}times \} larger in size, respectively. All of the source code and deep learning models described in this paper will be made available to the entire research community (https://github.com/aicentral/{FFNet}).},
	pages = {363--370},
	booktitle = {Image Analysis and Recognition},
	publisher = {Springer International Publishing},
	author = {Ibrahim, Ahmed and Abbott, A. Lynn and Hussein, Mohamed E.},
	editor = {Karray, Fakhri and Campilho, Aurélio and Cheriet, Farida},
	year = {2017},
	langid = {english},
	keywords = {Deep Branch, Deep Learning, Early Layer, Parallel Path, Text Detection}
}

Downloads: 0