The NTIS website and supporting ordering systems are undergoing a major upgrade from 8PM on September 25th through approximately October 6. During that time, much of the functionality, including subscription and product ordering, shipping, etc., will not be available. You may call NTIS at 1-800-553-6847 or (703) 605-6000 to place an order but you should expect delayed shipment. Please do NOT include credit card numbers in any email you might send NTIS.
Documents in the NTIS Technical Reports collection are the results of federally funded research. They are directly submitted to or collected by NTIS from Federal agencies for permanent accessibility to industry, academia and the public.  Before purchasing from NTIS, you may want to check for free access from (1) the issuing organization's website; (2) the U.S. Government Printing Office's Federal Digital System website http://www.gpo.gov/fdsys; (3) the federal government Internet portal USA.gov; or (4) a web search conducted using a commercial search engine such as http://www.google.com.
Accession Number ADA565180
Title Learning Hierarchical Feature Extractors for Image Recognition.
Publication Date Sep 2012
Media Count 197p
Personal Author Y. Boureau
Abstract Telling cow from sheep is effortless for most animals, but requires much engineering for computers. In this thesis, we seek to tease out basic principles that underlie many recent advances in image recognition. First, we recast many methods into a common unsupervised feature extraction framework based on an alternation of coding steps, which encode the input by comparing it with a collection of reference patterns, and pooling steps, which compute an aggregation statistic summarizing the codes within some region of interest of the image. Within that framework, we conduct extensive comparative evaluations of many coding or pooling operators proposed in the literature. Our results demonstrate a robust superiority of sparse coding (which decomposes an input as a linear combination of a few visual words) and max pooling (which summarizes a set of inputs by their maximum value). We also propose macrofeatures, which import into the popular spatial pyramid framework the joint encoding of nearby features commonly practiced in neural networks, and obtain significantly improved image recognition performance. Next, we analyze the statistical properties of max pooling that underlie its better performance through a simple theoretical model of feature activation. We then present results of experiments that confirm many predictions of the model. Beyond the pooling operator itself, an important parameter is the set of pools over which the summary statistic is computed. We propose locality in feature configuration space as a natural criterion for devising better pools. Finally, we propose ways to make coding faster and more powerful through fast convolutional feedforward architectures, and examine how to incorporate supervision into feature extraction schemes. Overall, our experiments offer insights into what makes current systems work so well, and state-of-the-art results on several image recognition benchmarks.
Keywords Feature extraction
Hierarchies
Image processing
Mathematical models
Recognition
Statistical analysis
Theses


 
Source Agency Non Paid ADAS
NTIS Subject Category 62 - Computers, Control & Information Theory
72E - Operations Research
72F - Statistical Analysis
Corporate Author New York Univ., NY. Dept. of Computer Science.
Document Type Thesis
Title Note Doctoral thesis.
NTIS Issue Number 1304
Contract Number N0014-09-1-0473

Science and Technology Highlights

See a sampling of the latest scientific, technical and engineering information from NTIS in the NTIS Technical Reports Newsletter

Acrobat Reader Mobile    Acrobat Reader