Browsing by Author "Burk, Thomas E."
Now showing 1 - 20 of 38
- Results Per Page
- Sort Options
Item A Spatial Semi-supervised Learning Method for Mining Multi-spectral Remote Sensing Imagery(2004-03-01) Vatsavai, Ranga R.; Shekhar, Shashi; Burk, Thomas E.Supervised learning, which is often used in land cover (thematic) classification of remote sensing imagery, has two limitations: first these techniques require large amounts of accurate training data to accurately estimate underlying statistical model parameters and secondly, the independent and identically distributed (i.i.d) assumptions made by these techniques do not hold true in the case of high-resolution satellite images. Recently, semi-supervised learning techniques that utilize large unlabeled training samples in conjunction with small labeled training data are becoming popular in machine learning, especially in text data mining. These techniques provide a viable solution to small training dataset problems; however, the techniques do not exploit spatial context. In this paper we explore methods that utilize unlabeled samples in supervised learning for classification of multi-spectral remote sensing imagery, while also taking into account the spatial context in the learning process. We extended the classical Expectation-Maximization (EM) technique to model spatial context via Markov Random Fields (MRF). We have conducted several experiments on real data sets and our classification procedure shows an improvement of 10% in overall classification accuracy. Further studies are necessary to assess the true potential and usefulness of this technique in varying geographic settings. Keywords: MAP, MLE, EM, Spatial Context, Auto-correlation, MRF, semi-supervised learning, mixture modelsItem Application of James-Stein and empirical Bayes procedures to simultaneous estimation problems in forest inventory(University of Minnesota, 1980-03-01) Burk, Thomas E.; Ek, Alan R.Traditional estimation procedures may ignore available auxilary information or use it only for regression, stratification, or in survey design. Such information, however, can be incorporated directly into estimation procedures. One case that has received much recent attention is where there exist K > 4 groups of simultaneous interest which are homogeneous (similar) with respect to their means. This paper describes two approaches (estimators) which incorporate auxilary information and thereby improve estimation efficiency as compared to commonly applied normal theory maximum likelihood estimators. Computer programs for implementation of the estimators are also described. Forest inventory problems provide an ideal application for such estimators. Simulation tests were conducted on four real forest populations covering volume and area estimation. Analysis of results showed consistent reductions in estimator total mean squared error. Confidence interval analyses are also presented.Item An approach to regional land cover classification in the Upper Great Lakes States.(University of Minnesota, 2000-07) Hansen, Sonja K.; Bolstad, Paul V.; Wilson, B. Tyler; Vatsavai, Ranga R.; Burk, Thomas E.; Bauer, Marvin E.Item Assessment of Carbon Flows Associated with Forest Management and Biomass Procurement for the Laskin Biomass Facility(University of Minnesota, 2008-11-22) Domke, Grant M.; Ek, Alan R.; Becker, Dennis R.; Espeleta, Javier F.; D’Amato, Anthony W.; Reich, Peter B.; Suh, Sangwon; Kilgore, Michael A.; Current, Dean A.; Hoganson, Howard M.; Burk, Thomas E.; Blinn, Charles R.This carbon life cycle analysis of forest-derived biomass was developed as part of a larger assessment by Minnesota Power detailing fuel supply, fuel procurement plans, and project engineering for a new 26-megawatt biomass generation facility in Hoyt Lakes, Minnesota. Forest-derived biomass is a renewable fuel that can be procured locally from forest harvest residues, mill residues, material from early thinnings and land cleaning, short rotation woody crops, brush, and urban wood waste. Energy generation from renewable fuels like forest biomass may dramatically alter the carbon balance in comparison to the use of fossil fuels like coal or natural gas. This study identifies the source and rate of carbon accumulation by tracking key inputs and outputs from forests through the conversion, regrowth and management activities over a 100-year period—the net carbon impact.Item Construction of a geographic information system for wildlife refuge planning : Agassiz National Wildlife Refuge.(University of Minnesota, 2002-02) Walker, Karen V.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Crab Orchard National Wildlife Refuge.(University of Minnesota, 2000-01) Sawaya, Kali E.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : DeSoto National Wildlife Refuge.(University of Minnesota, 1999-07) Littlefield, Shandy E.; Bolstad, Paul V.; Burk, Thomas E.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Mingo National Wildlife Refuge.(University of Minnesota, 2003-11) Geurts, Kari A.; Burk, Thomas E.; Bolstad, Paul V.; Balogh, MaryItem Construction of a geographic information system for wildlife refuge planning : Minnesota Valley National Wildlife Refuge.(University of Minnesota, 1999-07) Hansen, Sonja K.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Muscatatuck National Wildlife Refuge.(University of Minnesota, 2000-10) Kerns, Rick R.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Ottawa National Wildlife Refuge Complex.(University of Minnesota, 1999-07) Littlefield, Shandy E.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.; Hossein, AnwarItem Construction of a geographic information system for wildlife refuge planning : Patoka River National Wildlife Refuge.(University of Minnesota, 2000-05) Littlefield, Shandy E.; Bolstad, Paul V.; Burk, Thomas E.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Rice Lake National Wildlife Refuge.(University of Minnesota, 2004-05) Brown, Nicole L.; Burk, Thomas E.; Bolstad, Paul V.; Balogh, MaryItem Construction of a Geographic Information System for wildlife refuge planning : Seney National Wildlife Refuge.(University of Minnesota, 2003-10) Mueller, Brian D.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Shiawassee National Wildlife Refuge.(University of Minnesota, 1999-08) Hansen, Sonja K.; Bolstad, Paul V.; Burk, Thomas E.; Schomaker, John H.; Scharber, Michael A.Item Construction of a geographic information system for wildlife refuge planning : Squaw Creek National Wildlife Refuge.(University of Minnesota, 1999-10) Sawaya, Kali E.; Bolstad, Paul V.; Burk, Thomas E.; Schomaker, John H.Item Construction of a geographic information system for wildlife refuge planning : Swan Lake National Wildlife Refuge.(University of Minnesota, 1999-09) Kirk, Ryan W.; Burk, Thomas E.; Bolstad, Paul V.; Schomaker, John H.Item Construction of a Geographic Information System for Wildlife Refuge Planning: Agassiz National Wildlife Refuge(University of Minnesota, 2003-02) Walker, Karen V.; Burk, Thomas E.; Bolstad, Paul V.; Shomaker, John H.Item Description of a forest regulation simulator(University of Minnesota, 1980-03) Burk, Thomas E.; Rose, DietmarItem Development of a Model for Simulation of Forest Regulation Techniques(Minnesota Agricultural Experiment Station, 1980) Rose, Dietmar; Burk, Thomas E.