Browsing by Subject "Classification"

Now showing 1 - 20 of 20

Acquiring Bulk Compositions of Spinel Peridotite Xenoliths Using GIS and Digital Image Processing Techniques
(2022-12) Roberts, Amber
This study investigates the viability of using GIS techniques to estimate modal abundances of minerals in thin sections of peridotite and uses this data to calculate bulk chemical compositions of peridotitic mantle xenoliths from Oahu, Hawaii to determine their provenance. To answer these questions, I performed classifications using ArcGIS on nine xenolith samples to determine their modal abundances, which were then used to calculate their bulk chemical compositions. My results show that GIS has the potential to be a useful tool for non-destructive analysis of modal abundances. They also support a role for melt-rock reaction occurring between migrating melts and peridotite in the oceanic lithospheric mantle, resulting in the production of dunites. This study is a first step in utilizing GIS to assist with thin section analyses and fills a gap in existing chemical data for Hawaiian mantle xenoliths.
Algorithms for Semisupervised learning on graphs
(2018-12) Flores, Mauricio
Laplacian regularization has been extensively used in a wide variety of semi-supervised learning tasks over the past fifteen years. In recent years, limitations of the Laplacian regularization have been exposed, leading to the development of a general class of Lp-based Laplacian regularization models. We propose novel algorithms to solve the resulting optimization problem, as the amount of unlabeled data increases to infinity, while the amount of labeled data remains fixed and is very small. We explore a practical application to recommender systems.
Automated segmentation and pathology detection in ophthalmic images
(2014-07) Roy Chowdhury, Sohini
Computer-aided medical diagnostic system design is an emerging inter-disciplinary technology that assists medical practitioners for providing quick and accurate diagnosis and prognosis of pathology. Since manual assessments of digital medical images can be both ambiguous and time-consuming, computer-aided image analysis and evaluation systems can be beneficial for baseline diagnosis, screening and disease prioritization tasks. This thesis presents automated algorithms for detecting ophthalmic pathologies pertaining to the human retina that may lead to acquired blindness in the absence of timely diagnosis and treatment. Multi-modal automated segmentation and detection algorithms for diabetic manifestations such as Diabetic Retinopathy and Diabetic Macular Edema are presented. Also, segmentation algorithms are presented that can be useful for automated detection of Glaucoma, Macular Degeneration and Vein Occlusions. These algorithms are robust to normal and pathological images and incur low computationally complexity.First, we present a novel blood vessel segmentation algorithm using fundus images that extracts the major blood vessels by applying high-pass filtering and morphological transforms followed by addition of fine vessel pixels that are classified by a Gaussian Mixture Model (GMM) classifier. The proposed algorithm achieves more than 95% vessel segmentation accuracy on three publicly available data sets. Next, we present an iterative blood vessel segmentation algorithm that initially estimates the major blood vessels, followed by iterative addition of fine blood vessel segments till a novel stopping criterion terminates the iterative vessel addition process. This iterative algorithm is specifically robust to thresholds since it achieves 95.35% vessel segmentation accuracy with 0.9638 area under ROC curve (AUC) on abnormal retinal images from the publicly available STARE data set.We propose a novel rule-based automated optic disc (OD) segmentation algorithm that detects the OD boundary and the location of vessel origin (VO) pixel. This algorithm initially detects OD candidate regions at the intersection of the bright regions and the blood vessels in a fundus image subjected to certain structural constraints, followed by the estimation of a best fit ellipse around the convex hull that combines all the detected OD candidate regions. The centroid of the blood vessels within the segmented OD boundary is detected as the VO pixel location. The proposed algorithm results in an average of 80% overlap score on images from five public data sets.We present a novel computer-aided screening system (DREAM) that analyzes fundus images with varying illumination and fields of view, and generates a severity grade for non-proliferative diabetic retinopathy (NPDR) using machine learning. Initially, the blood vessel regions and the OD region are detected and masked as the fundus image background. Abnormal foreground regions corresponding to bright and red retinopathy lesions are then detected. A novel two-step hierarchical classification approach is proposed where the non-lesions or false positives are rejected in the first step. In the second step, the bright lesions are classified as hard exudates and cotton wool spots, and the red lesions are classified as hemorrhages and micro-aneurysms. Finally, the number of lesions detected per image is combined to generate a severity grade. The DReAM system achieves 100% sensitivity, 53.16% specificity and 0.904 AUC on a publicly available MESSIDOR data set with 1200 images. Additionally, we propose algorithms that detect post-operative laser scars and fibrosed tissues and neovascularization in fundus images. The proposed algorithm achieves 94.74% sensitivity and 92.11% specificity for screening normal images in the STARE data set from the images with proliferative diabetic retinopathy (PDR). Finally, we present a novel automated system that segments six sub-retinal thickness maps from optical coherence tomography (OCT) image stacks of healthy patients and patients with diabetic macular edema (DME). First, each image in the OCT stack is denoised using a Wiener Deconvolution algorithm that estimates the speckle noise variance using a Fourier-domain based structural error. Next, the denoised images are subjected to an iterative multi-resolution high-pass filtering algorithm that detects seven sub-retinal surfaces in six iterative steps. The thicknesses of each sub-retinal layer for all scans from a particular OCT stack are then combined to generate sub-retinal thickness maps. Using the proposed system the average inner sub-retinal layer thickness in abnormal images is estimated as 275 um (r = 0.92) with an average error of 9.3 um, while the average thickness of the outer segments in abnormal images is estimated as 57.4 um (r = 0.74) with an average error of 3.5 um. Further analysis of the thickness maps from abnormal OCT image stacks demonstrates irregular plateau regions in the inner nuclear layer (INL) and outer nuclear layer (ONL), whose area can be estimated with r = 0.99 by the proposed segmentation system.
Characterization and formation of high-chroma features in loamy soils of southern Minnesota
(2012-12) Pribyl, Douglas Wayne
High‐chroma features have not been adequately defined under existing terminology or classified under existing systems. The terms “masses” as a subclass of concentrations used in field definitions and “loose infillings” used in micromorpological classifications come closest but are not fully satisfactory. Defined descriptively, high‐chroma features have a typical color of 7.5YR 5/8, are usually less than 1 to 2 mm in diameter, are poorly cemented, and have a sharp external boundary with the soil matrix. They are found in well‐drained to poorly drained soils with first‐appearance typically at depths of 50 to 100 cm. A study was undertaken to more fully characterize and classify high‐chroma features and to provide more accurate interpretations of feature morphology for applications in environmental and soil quality, plant nutrition, and soil genesis. High‐chroma features found within peds having varying degrees of hydromorphic expression were assigned to classes depending on internal color and color patterns. Material removed from features, halos, and the soil matrix was analyzed using a low‐power stereomicroscope, SEM/EDS, TEM/ED, μ‐XRD, ICP, and stain tests to determine properties and composition. Four formation hypotheses are proposed: (1) a non‐pedogenic origin, features having developed from the weathering of an inherited precursor mineral; (2) a pedogenic origin resulting from the formation and infilling of vesicles that formed at depth shortly after deglaciation but are no longer actively forming; (3) a pedogenic origin but features are actively forming; (4) formation by dissolution of a soluble mineral fragment and subsequent infilling of the resulting void, analogous to the formation of a geode. Although high‐chroma features might develop by more than one pathway, a non‐pedogenic origin is favored. Non‐pedogenic hypothesis (1) and the hybrid geodic hypothesis (4) offer the most efficient explanations for the presence of silt, iron, and manganese within high‐chroma features. A proposed weathering sequence based on feature classification and evidence for the presence of manganese nodules in the till‐source bedrock also support a nonpedogenic origin. Pedogenic hypotheses require a sequence of events of uncertain and in some cases seemingly low probability. Existing classification systems offer little insight into genesis. Most importantly, given the evidence for a non‐pedogenic origin, high‐chroma features should not be interpreted or classified as redoximorphic features as the term is typically used in the field. Although high‐chroma features may result from alternating periods of oxidation and reduction, when used alone they are ambiguous indicators of seasonal wetness.
A complete system for polyps flagging in virtual colonoscopy
(University of Minnesota. Institute for Mathematics and Its Applications, 2011-04) Fiori, Marcelo; Musé, Pablo; Sapiro, Guillermo
Computational Techniques to Identify Rare Events in Spatio-temporal Data
(2018-05) Mithal, Varun
Recent attention on the potential impacts of land cover changes to the environment as well as long-term climate change has increased the focus on automated tools for global-scale land surface monitoring. Advancements in remote sensing and data collection technologies have produced large earth science data sets that can now be used to build such tools. However, new data mining methods are needed to address the unique characteristics of earth science data and problems. In this dissertation, we explore two of these interesting problems, which are (1) build predictive models to identify rare classes when high quality annotated training samples are not available, and (2) classification enhancement of existing imperfect classification maps using physics-guided constraints. We study the problem of identifying land cover changes such as forest fires as a supervised binary classification task with the following characteristics: (i) instead of true labels only imperfect labels are available for training samples. These imperfect labels can be quite poor approximation of the true labels and thus may have little utility in practice. (ii) the imperfect labels are available for all instances (not just the training samples). (iii) the target class is a very small fraction of the total number of samples (traditionally referred to as the rare class problem). In our approach, we focus on leveraging imperfect labels and show how they, in conjunction with attributes associated with instances, open up exciting opportunities for performing rare class prediction. We applied this approach to identify burned areas using data from earth observing satellites, and have produced a database, which is more reliable and comprehensive (three times more burned area in tropical forests) compared to the state-of-art NASA product. We explore approaches to reduce errors in remote sensing based classification products, which are common due to poor data quality (eg., instrument failure, atmospheric interference) as well as limitations of the classification models. We present classification enhancement approaches, which aim to improve the input (imperfect) classification by using some implicit physics-based constraints related to the phenomena under consideration. Specifically, our approach can be applied in domains where (i) physical properties can be used to correct the imperfections in the initial classification products, and (ii) if clean labels are available, they can be used to construct the physical properties.
Data mining techniques for enhancing protein function prediction
(2010-04) Pandey, Gaurav
Proteins are the most essential and versatile macromolecules of life, and the knowledge of their functions is crucial for obtaining a basic understanding of the cellular processes operating in an organism as well as for important applications in biotechnology, such as the development of new drugs, better crops, and synthetic biochemicals such as biofuels. Recent revolutions in biotechnology has given us numerous high-throughput experimental technologies that generate very useful data, such as gene expression and protein interaction data, that provide high-resolution snapshots of complex cellular processes and a novel avenue to understand their underlying mechanisms. In particular, several computational approaches based on the principle of Guilt by Association (GBA) have been proposed to predict the function(s) of the protein are inferred from those of other proteins that are "associated" to it in these data sets. In this thesis, we have developed several novel methods for improving the performance of these approaches by making use of the unutilized and under-utilized information in genomic data sets, as well as their associated knowledge bases. In particular, we have developed pre-processing methods for handling data quality issues with gene expression (microarray) data sets and protein interaction networks that aim to enhance the utility of these data sets for protein function prediction. We have also developed a method for incorporating the inter-relationships between functional classes, as captured by the ontologies in Gene Ontology, into classification-based protein function prediction algoriths, which enabled us to improve the quality of predictions made for several functional classes, particularly those with very few member proteins (rare classes). Finally, we have developed a novel association analysis-based biclustering algorithm to address two major challenges with traditional biclustering algorithms, namely an exhaustive search of all valid biclusters satisfying the definition specified by the algorithm, and the ability to search for small biclusters. This algorithm makes it possible to discover smaller sized biclusters that are more significantly enriched with specific GO terms than those produced by the traditional biclustering algorithms. Overall, the methods proposed in this thesis are expected to help uncover the functions of several unannotated proteins (or genes), as shown by specific examples cited in some of the chapters. To conclude, we also suggest several opportunities for further progress on the very important problem of protein function prediction
Enhancing Machine Learning Classification for Electrical Time Series with Additional Domain Applications
(2019-11) Valovage, Mark
Recent advances in machine learning have significant, far-reaching potential in electrical time series applications. However, many methods cannot currently be implemented in real world applications due to multiple challenges. This thesis explores solutions to many of these challenges in an effort to realize the full potential of applying machine learning to dynamic electrical systems. This thesis focuses on two areas: electricity disaggregation and time series shapelets. However, the contributions below can be applied to dozens of other domains. Electricity disaggregation identifies individual appliances from one or more aggregate data streams. In first world countries, disaggregation has the potential to eliminate billions of dollars of waste each year, while in developing countries, disaggregation could reduce costs enough to help provide electricity to over a billion people who currently have no access to it. Existing disaggregation methods cannot be applied to real-world households because they are too sensitive to varying noise levels, require parameters to be tuned to individual houses or appliances, make incorrect assumptions about real-world data, or are too resource intensive for inexpensive hardware. This thesis details label correction, a process to automatically correct user-labeled training samples, to increase classification accuracy. It also details an approach to unsupervised learning that is scalable to hundreds of millions of buildings using two novel approaches: event detection without parameter tuning and iterative discovery without appliance models. Time series shapelets are small subsequences of time series used for classification of unlabeled time series. While shapelets can be used for electricity disaggregation, they have applications to dozens of other domains. However, little research has been done on the distance metric used by shapelets. This distance metric is critical, as it is the sole feature a shapelet uses to discriminate between samples from different classes. This thesis details two contributions to time series shapelets. The first, selective z-normalization, is a technique that increases the shapelet classification accuracy by discovering a combination of z-normalized and non-normalized shapelets. The second is computing shapelet-specific distances, a technique to increase accuracy by finding a unique distance metric for each shapelet.
Global self-similarity and saliency measures based on sparse representations for classification of objects and spatio-temporal sequences
(2012-12) Somasundaram, Guruprasad
Extracting the truly salient regions in images is critical for many computer vision applications. Salient regions are considered the most informative regions of an image. Traditionally these salient regions have always been considered as local phenomena in which the salient regions stand out as local extrema with respect to their immediate neighbors. We introduce a novel global saliency metric based on sparse representation in which the regions that are most dissimilar with respect to the entire image are deemed salient. We examine our definition of saliency from the theoretical stand point of sparse representation and minimum description length. Encouraged by the efficacy of our method in modeling foreground objects, we propose two classification methods for recognizing objects in images. First, we introduce two novel global self-similarity descriptors for object representation which can directly be used in any classification framework. Next, we use our salient feature detection approach with conventional region descriptors in a bag-of-features framework. Experimentally we show that our feature detection method enhances the bag-of-features framework. Finally, we extend our salient bag-of-features approach to the spatio-temporal domain for use with three-dimensional dense descriptors. We apply this method successfully to video sequences involving human actions. We obtain state-of-the-art recognition rates in three distinct datasets involving sports and movie actions.
Hierarchical dictionary learning for invariant classification
(University of Minnesota. Institute for Mathematics and Its Applications, 2009-09) Bar, Leah; Sapiro, Guillermo
Incorporating biological knowledge of genes into microarry data analysis.
(2009-04) Tai, Feng
Microarray data analysis has become one of the most active research areas in bioinformatics in the past twenty years. An important application of microarray technology is to reveal relationships between gene expression profiles and various clinical phenotypes. A major characteristic in microarray data analysis is the so called "large p, small n" problem, which makes it difficult for parameter estimation. Most of the traditional statistical methods developed in this area target to overcome this difficulty. The most popular technique is to utilize an L1 norm penalty to introduce sparsity into the model. However, most of those traditional statistical methods for microarray data analysis treat all genes equally, as for usual covariates. Recent development in gene functional studies have revealed complicated relationships among genes from biological perspectives. Genes can be categorized into biological functional groups or pathways. Such biological knowledge of genes along with microarray gene expression profiles provides us the information of relationships not only between gene and clinical outcomes but also among the genes. Utilizing such information could potentially improve the predictive power and gene selection. The importance of incorporating biological knowledge into analysis has been increasingly recognized in recent years and several new methods have been developed. In our study, we focus on incorporating biological information, such as the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, into microarray data analysis for the purpose of prediction. Our first method aims implement this idea by specifying different L1 penalty terms for different gene functional groups. Our second method models a covariance matrix for the genes by assuming stronger within-group correlations and weaker between-group correlations. The third method models spatial correlations among the genes over a gene network in a Bayesian framework.
Left-handed BCI - examining effects of handedness and hand dominance on EEG grasp classification
(2022-12) Dowling, Dale
Brain-Computer Interfaces (BCIs) are of high potential use to individuals whosemotor function is impaired, or who have undergone a loss of limb or limb functionality. Electroencephalography (EEG) is one popular method of collecting signals from the brain, and is commonly used in cases where other sensing methods are difficult or impossible. This method of collecting brain signal data has been shown, when used in conjunction with Electromyography (EMG) data, to be capable of classifying fine hand movements with a high degree of accuracy, providing avenues for the design of highly attenuated prosthetic limbs. Studies which have examined such uses for BCIs, however, seldom examine the effects of handedness, as well as off-hand or dualhanded motion, on classification accuracy. This study examines the effects of hand use and hand dominance on the performance of several classifiers derived from EEG data. Data was collected, using the OpenBCI EEG Electrode Cap Kit, for 16 participants (9 right-handed, 7 left-handed), on a set of 6 grasp types, and a selection of 5 classification algorithms (Naive Bayes, Decision Tree, Logistic Regression, Support Vector Machine, and Neural Network) commonly found in previous works were used. Outcomes of the study indicate that Neural Networks are best suited among these classifiers to determine hand motion in a dual-handed environment, and that, while providing hand-dominance data for classification training may not improve accuracy in all cases, design and feature changes based on factors such as hand-dominance may improve the performance of BCIs based on EEG data.
Model-based methods for high-dimensional multivariate analysis
(2017-04) Molstad, Aaron
This thesis consists of three main parts. In the first part, we propose a penalized likelihood method to fit the linear discriminant analysis model when the predictor is matrix valued. We simultaneously estimate the means and the precision matrix, which we assume has a Kronecker product decomposition. Our penalties encourage pairs of response category mean matrix estimators to have equal entries and also encourage zeros in the precision matrix estimator. To compute our estimators, we use a blockwise coordinate descent algorithm. To update the optimization variables corresponding to response category mean matrices, we use an alternating minimization algorithm that takes advantage of the Kronecker structure of the precision matrix. We show that our method can outperform relevant competitors in classification, even when our modeling assumptions are violated. We analyze an EEG dataset to demonstrate our method's interpretability and classification accuracy. In the second part, we propose a class of estimators of the multivariate response linear regression coefficient matrix that exploits the assumption that the response and predictors have a joint multivariate normal distribution. This allows us to indirectly estimate the regression coefficient matrix through shrinkage estimation of the parameters of the inverse regression, or the conditional distribution of the predictors given the responses. We establish a convergence rate bound for estimators in our class and we study two examples, which respectively assume that the inverse regression's coefficient matrix is sparse and rank deficient. These estimators do not require that the forward regression coefficient matrix is sparse or has small Frobenius norm. Using simulation studies, we show that our estimators outperform competitors. In the final part of this thesis, we propose a framework to shrink a user-specified characteristic of a precision matrix estimator that is needed to fit a predictive model. Estimators in our framework minimize the Gaussian negative log-likelihood plus an L1 penalty on a linear or affine function evaluated at the optimization variable corresponding to the precision matrix. We establish convergence rate bounds for these estimators and we propose an alternating direction method of multipliers algorithm for their computation. Our simulation studies show that our estimators can perform better than competitors when they are used to fit predictive models. In particular, we illustrate cases where our precision matrix estimators perform worse at estimating the population precision matrix while performing better at prediction.
Modern Classification with Big Data
(2018-07) Wang, Boxiang
Rapid advances in information technologies have ushered in the era of "big data" and revolutionized the scientific research across many disciplines, including economics, genomics, neuroscience, and modern commerce. Big data creates golden opportunities but has also arisen unprecedented challenges due to the massive size and complex structure of the data. Among many tasks in statistics and machine learning, classification has diverse applications, ranging from improving daily life to reaching the new frontiers of science and engineering. This thesis will discuss the envisions of broader approaches to modern classification methodologies, as well as computational considerations to cope with the big data challenges. Chapter 2 of the thesis presents a modern classification method named data-driven generalized distance weighted discrimination. A fast algorithm with an emphasis on computational efficiency for big data will be introduced. Our method is formulated in a reproducing kernel Hilbert space, and learning theory of the Bayes risk consistency will be developed. We will use extensive benchmark data applications to demonstrate that the prediction accuracy of our method is highly competitive with state-of-the-art classification methods including support vector machine, random forest, gradient boosting, and deep neural network. Chapter 3 introduces sparse penalized DWD for high-dimensional classification, which is commonly used in the era of big data. We develop a very efficient algorithm to compute the solution path of the sparse DWD at a given fine grid of regularization parameters. Chapter 4 proposes multicategory kernel distance weighted discrimination for multi-class classification. The proposal is defined as a margin-vector optimization problem in a reproducing kernel Hilbert space. This formulation is shown to enjoy Fisher consistency. We develop an accelerated projected gradient descent algorithm to fit multicategory kernel DWD. Chapter 5 develops a magic formula for doing CV in the context of large margin classification. We design a novel and successful algorithm to fit and tune the support vector machine.
Novel Biomarker Identification Approaches for Schizophrenia using fMRI and Retinal Electrophysiology
(2017-11) Moghimi, Pantea
Schizophrenia is a chronic mental illness. The exact cause if schizophrenia is not yet known. Extensive research has been done to identify robust biomarkers for the disease using non-invasive brain imaging techniques. A robust biomarker can be informative about pathophysiology of the disease and can guide clinicians into developing more effective interventions. The aim of this dissertation is two folds. First, we seek to identify robust biomarkers using resting state fMRI activity from a cohort of schizophrenic and healthy subjects in a purely data driven approach. We will calculate multivariate network measures and use them as features for classification of the subjects into healthy and diseased. The network measures will be calculated using nodes defined by the AAL anatomical atlas as well as a functional atlas constructed from the fMRI activity. Network measures with high classification rate may be used as potential biomarkers. We will employ double cross-validation to estimate generalizability of our results to a new population of subjects that were not used in biomarker identification. Second, we seek to identify biomarkers using electroretinogram (ERG). We will use a data driven approach to classify individuals based on the pattern of retinal activity they exhibit in response to visual stimulation. Characteristics of the ERG result in high classification rate are presented as potential biomarkers of schizophrenia.
Sharing the Load - Offloading Processing and Improving Emotion Classification for the SoftBank Robot Pepper""
(2021-04) Savela, Shawn
Pepper is a humanoid robot created by SoftBank Robotics that was designed and built with the purpose of being used for robot-human interaction. There is an application interface that allows development of custom interactive programs as well as a number of built-in applications that can be extended and used when creating other custom programs for the robot. Among the pre-installed applications are applications that will classify a person's emotion and mood using data from several data points including facial characteristics and vocal pitch and tone. Due to the Covid-19 pandemic many people have been wearing face masks in both public and private areas. Detecting emotions based on facial recognition and voice tone analysis may not be as accurate when a person is wearing a mask. An alternative method that can be used to classify emotion is to analyze the actual words that are spoken by a person. However, this feature is not currently available on Pepper. In this study we describe a software solution that will allow Pepper to perform sentiment classification based on spoken words using a neural network. We will describe the testing procedure that was used to interview participants by Pepper and compare the F1 score of each classification method with each other. Pepper was able to be programmed to use a neural network for emotion classification. A total of 32 participants were interviewed, with the NLP spoken-word analysis classification achieving an averaged F1 score of .2860 as compared to the built-in software average F1 scores of .2362 from the mood application, .1986 from the vocal tone and pitch application, and .0811 from the facial characteristics application.
Sparse coding and dictionary learning based on the MDL principle
(University of Minnesota. Institute for Mathematics and Its Applications, 2010-10) Ramírez, Ignacio; Sapiro, Guillermo
Textural Analysis and Substrate Classification in the Nearshore Region of Lake Superior Using High-Resolution Multibeam Bathymetry
(2017-09) Dennison, Andrew
Classification of the seafloor substrate can be done with a variety of methods. These methods include Visual (dives, drop cameras); mechanical (cores, grab samples); acoustic (statistical analysis of echosounder returns). Acoustic methods offer a more powerful and efficient means of collecting useful information about the bottom type. Due to the nature of an acoustic survey, larger areas can be sampled, and by combining the collected data with visual and mechanical survey methods provide greater confidence in the classification of a mapped region. During a multibeam sonar survey, both bathymetric and backscatter data is collected. It is well documented that the statistical characteristic of a sonar backscatter mosaic is dependent on bottom type. While classifying the bottom-type on the basis on backscatter alone can accurately predict and map bottom-type, i.e a muddy area from a rocky area, it lacks the ability to resolve and capture fine textural details, an important factor in many habitat mapping studies. Statistical processing of high-resolution multibeam data can capture the pertinent details about the bottom-type that are rich in textural information. Further multivariate statistical processing can then isolate characteristic features, and provide the basis for an accurate classification scheme. The development of a new classification method is described here. It is based upon the analysis of textural features in conjunction with ground truth sampling. The processing and classification result of two geologically distinct areas in nearshore regions of Lake Superior; off the Lester River,MN and Amnicon River, WI are presented here, using the Minnesota Supercomputer Institute's Mesabi computing cluster for initial processing. Processed data is then calibrated using ground truth samples to conduct an accuracy assessment of the surveyed areas. From analysis of high-resolution bathymetry data collected at both survey sites is was possible to successfully calculate a series of measures that describe textural information about the lake floor. Further processing suggests that the features calculated capture a significant amount of statistical information about the lake floor terrain as well. Two sources of error, an anomalous heave and refraction error significantly deteriorated the quality of the processed data and resulting validate results. Ground truth samples used to validate the classification methods utilized for both survey sites, however, resulted in accuracy values ranging from 5 -30 percent at the Amnicon River, and between 60-70 percent for the Lester River. The final results suggest that this new processing methodology does adequately capture textural information about the lake floor and does provide an acceptable classification in the absence of significant data quality issues.
Wetland Inventory and Classification for Carlton and South St. Louis Counties : Final Report and Deliverables
(University of Minnesota Duluth, 2008-12-31) Host, George E; Meysembourg, Paul
Accurate maps of the type and locations of wetlands are critical for land use planning, particularly for watersheds undergoing rapid develoment or facing increased development pressure. The important role wetlands play in maintaining habitat, water quality and surface and ground-water protection is well documented, but cun*ent information on the types, sizes, and locations of wetlands is difficult to obtain. As coastal environments come under increased pressure from development, this infonnation is essential for zoning, buildout scenarios and numerous other planning objectives. Within the Coastal Program boundary, however，up-to-date information on wetland type and distribution is sparse, outdated, or lacking for many watersheds. While the National Wetland Inventory is the most extensive and commonly used inventory, the limitations with respect to spatial and classification accuracy are well-recogiiized. Over several iterations, we have systematically been mapping wetlands within high- gi*owth areas of the Minnesota's Lake Superior Coastal Program. The objective of the current proposal is to use recent MN DNR aerial photography and other spatial data to delineate and characterize wetlands for the southwestern portion the Coastal Program area. These includes approximately tliree townships in Carlton County and watershed extensions into St. Louis County (Figure 1). Our primary end products are digital maps of classified wetlands and with associated data tables, which are here provided to the Lake Superior Coastal Program for distribution to decision makers and the general public. Wetland maps are delivered in two fomiats. As part of this final report to the MN DNR, we have included a DVD that contains the rectified raw imagery, inteipreted wetland in GIS fonnat, and metadata for the data layers. We have also created, as part of the CoastalGIS website at the Natural Resources Research Institute, downloadable and online versions of the interpreted wetlands. The download versions are delivered in ESRI shapefile fonnat, with associated metadata. We also provide an interactive version using the Arc Internet Map Server, which allows maps to be viewed and manipulated over the Internet with a standard web brower. The NOAA-funded CoastalGIS web site was established in March 2002 to sei*ve as a clearinghouse for spatial data relevant to the Coastal Program. The site currently contains a wide range of data sets on natural resources and infrastructure，and is designed to assist local decision makers and the general public in land use planning. The CoastalGIS web site can be accessed at: http://www. nrri. umn. edu/Coastal GIS
Wetlands of Cook Inlet Basin, Alaska: Classification and Contributions to Stream Flow
(2017-04) Gracz, Michael
Wetlands face threats from global change, even as protections have been institutionalized to conserve the amenities they provide. These institutional protections frequently rely on a wetland classification system to guide conservation. In the Cook Inlet Basin of Alaska, USA (CIB), for example, best wetland assessment practices require the use of a classification system to ensure the conservation of the most valuable amenities. However, the systems used widely in the USA outside of Alaska, where peatlands are not common, inadequately describe the diversity of peatlands on the glaciated landscape of the CIB. Here I present a new Cook Inlet Classification system (CIC) organized around the hydrogeologic settings of wetlands in the CIB. The variables most strongly correlated with ecological differences within major geomorphic classes were used to construct a system supported by ample field data. The CIC produced greater within-class similarity than other widely-used systems, likely due to the overriding importance of the seasonal variability of water levels in CIB peatlands. The CIC has been mapped over an area of 7600 km2 and has guided wetland functional assessments in the CIB, and may be adaptable to any region supporting peatlands on glacial landforms. The harmful effects of a warming climate on aquatic resources may be partially ameliorated by discharge of shallow groundwater from peatlands to streams. This potential benefit of peatlands was investigated in the CIB using end-member mixing analysis (EMMA) and a sensitivity analysis of a water budget to quantify the contribution from extensive peatlands formed over glacial lake deposits to stream flow during the dry-season. Although peatlands in this hydrogeologic setting are common globally, the discharge from them is challenging to quantify. A spatially distributed sampling protocol at a single point-in-time produced a reliable EMMA showing that over half of stream flow on a day during the summer dry period originated near the surface of peatlands. This finding is being used to establish the value of peatlands for buffering increases in stream temperature, which have exceeded tolerances of commercially important fishes in the CIB. The analysis also suggests that differences in hydrogeologic setting influence shallow groundwater hydrology in peatlands.

University Digital Conservancy

Browse by Subject

Browsing by Subject "Classification"