APSIPA Transactions on Signal and Information Processing > Vol 4 > Issue 1

The rationale for ensemble and meta-algorithmic architectures in signal and information processing

Industrial Technology Advances

Steven J. Simske, HP, USA, steven.simske@hp.com
 
Suggested Citation
Steven J. Simske (2015), "The rationale for ensemble and meta-algorithmic architectures in signal and information processing", APSIPA Transactions on Signal and Information Processing: Vol. 4: No. 1, e8. http://dx.doi.org/10.1017/ATSIP.2015.10

Publication Date: 02 Sep 2015
© 2015 Steven J. Simske
 
Subjects
 
Keywords
Machine learningCombinational approachesMeta-algorithmicsEnsemble approachesAlgorithms
 

Share

Open Access

This is published under the terms of the Creative Commons Attribution licence.

Downloaded: 1025 times

In this article:
I. INTRODUCTION 
II. ENSEMBLE ARCHITECTURES 
III. META-ALGORITHMIC ARCHITECTURES 
IV. A NEW MODEL FOR ARCHITECTURE 
V. BIOLOGICAL SIGNAL PROCESSING AND META-ALGORITHMS 
VI. IMAGE PROCESSING, OBJECT SEGMENTATION AND CLASSIFICATION, AND META-ALGORITHMS 
VII. DISCUSSION, CONCLUSIONS, AND THE FUTURE 

Abstract

We are living through an historic era in computing. As the price of data storage and processing continues to plummet, we are moving closer to a world where exhaustive search makes sense for certain types of intelligent systems. Signal and image processing are two related domains that benefit from this ubiquity of data storage and computing power. In this paper, a new, more collaborative, approach to solving signal and image processing tasks is built from the ground up to take into account the reality of this new age of data and computing superfluity. Starting with the mature field of ensemble methods and moving to the more-recently introduced field of meta-algorithmics, systems can be designed which are by nature to specifically incorporate new machine-learning technologies. These are more robust, more accurate, more adaptive, and ultimately less costly to build and maintain than the traditional machine-learning approaches. Applications to image and signal processing will then be discussed. Combined, these examples illustrate a new meta-architectural approach to the creation of machine intelligence systems.

DOI:10.1017/ATSIP.2015.10