Data Mining - Concepts and Techniques.

Statistics for Spatial Data.

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

Artificial neural networks (ANNs) constitute a class of flexible nonlinear models designed to mimic biological neural systems. In this entry, we introduce ANN using familiar econometric terminology and provide an overview of ANN modeling approach and its implementation methods. † Correspondence: Chung-Ming Kuan, Institute of Economics, Academia Sinica, 128 Academia Road, Sec. 2, Taipei 115, Taiwan; ckuan@econ.sinica.edu.tw. †† I would like to express my sincere gratitude to the editor, Professor Steven Durlauf, for his patience and constructive comments on early drafts of this entry. I also thank Shih-Hsun Hsu and Yu-Lieh Huang for very helpful suggestions. The remaining errors are all mine.

Artificial neural networks

There is a large class of modeling problems where the complexity of the underlying phenomena is overwhelming and hence the accuracy of mathematical models is limited. Our approach to this class of problems is to design frameworks that bring together physically based and data-driven models, and incorporate the tacit knowledge of experts by providing visual exploration and feedback capabilities. This paper presents such a novel computer-assisted framework for accurate geospatial modeling applied to improve groundwater recharge and discharge (R/D) patterns. The novelty of our work is in designing a methodology for ranking and extracting relationships, as well as in developing a general framework for building accurate geospatial models. The framework combines variables derived using physically based inverse modeling with auxiliary geospatial variables directly sensed, ranks variables and extracts variable relationships using data-driven (“machine learning”) techniques, and supports partially expert-driven trial-and-error experimentation and more rigorous optimization, as well as visual explorations, to derive more accurate models for R/D pattern estimation. When the framework was tested by experts, it led to a high level of consistency between the machine-learning-based knowledge and the experts9 knowledge about R/D distribution. The prototype solution of the framework is available for downloading at http://isda.ncsa.uiuc.edu/Sp2Learn/.

/pdf/a-framework-for-accurate-geospatial-modeling-using-image-3oy75maou0.pdf

A framework for accurate geospatial modeling using image ranking and machine learning

Scanning electron microscopy with energy-dispersive X-ray spectrometry (SEM-EDS) was used to characterize a macroscopic artifact from Peru (a Moche cast figure) [1]. A 3D hyperspectral datacube was acquired over 51.2 hours, measuring 11,520 x 9,984 pixels with 2,048 energy channels per pixel; each pixel corresponds to 900 nm in the x and y dimensions. With current technology, interacting with this hyperspectral dataset to perform exploratory analysis, displaying gigapixel elemental maps, and surveying terabyte-scale spectral data is problematic.

/pdf/interactive-analysis-of-terabyte-sized-sem-eds-hyperspectral-2ex9lio7qc.pdf

Interactive Analysis of Terabyte-sized SEM-EDS Hyperspectral Images

/pdf/impact-of-sampling-and-augmentation-on-generalization-59nctazxf1.pdf

Impact of Sampling and Augmentation on Generalization Accuracy of Microscopy Image Segmentation Methods | NIST

GeoLearn: An exploratory framework for extracting information and knowledge from remote sensing imagery

This paper presents a theoretical analysis of error calculations during a system design. A system is understood as a measurement chain that consists of units called transducers. In general, transducers are connected in three basic configurations; serial, parallel and with feedback. In order to illustrate the analysis of error calculations we consider only two connected units for each configuration of a measurement chain. In this paper, we analyze two types of error calculations for these three basic configurations. First, a standard error calculation is described assuming that transducer errors in a measurement chain are mutually correlated. System designers frequently assume an error correlation and therefore use standard error calculations. Second, error calculations are performed assuming that transducer errors are not mutually correlated. The case of mutually uncorrelated transducer errors is very common in a real system design since the numerical specifications of individual transducers are obtained from multiple independent catalogs. Thus, it is the uncorrelated nature of transducer errors that requires a modification of standard error calculations. We analyze error calculations for each configuration of transducers and for mutually correlated and uncorrelated transducer errors. In conclusion, the error formulas assuming mutually uncorrelated transducer errors model a real system design more accurately than the standard error formulas.

/pdf/resulting-errors-of-measurement-chains-jh2e5pn2xi.pdf

Peter Bajcsy

Papers

A framework for accurate geospatial modeling using image ranking and machine learning

Interactive Analysis of Terabyte-sized SEM-EDS Hyperspectral Images

Impact of Sampling and Augmentation on Generalization Accuracy of Microscopy Image Segmentation Methods | NIST

GeoLearn: An exploratory framework for extracting information and knowledge from remote sensing imagery

Resulting errors of measurement chains