Active Learning of Very-High Resolution Optical Imagery with SVM: Entropy vs Margin Sampling

An active learning method is proposed for the semi-automatic selection of training sets in remote sensing image classification. The method adds iteratively to the current training set the unlabeled pixels for which the prediction of an ensemble of classifiers based on bagged training sets show maximum entropy. This way, the algorithm selects the pixels that are the most uncertain and that will improve the model if added in the training set. The user is asked to label such pixels at each iteration. Experiments using support vector machines (SVM) on an 8 classes QuickBird image show the excellent performances of the methods, that equals accuracies of both a model trained with ten times more pixels and a model whose training set has been built using a state-of-the-art SVM specific active learning method

Limited-sampling strategy models for estimating the pharmacokinetic parameters of 4-methylaminoantipyrine, an active metabolite of dipyrone

Bioanalytical data from a bioequivalence study were used to develop limited-sampling strategy (LSS) models for estimating the area under the plasma concentration versus time curve (AUC) and the peak plasma concentration (Cmax) of 4-methylaminoantipyrine (MAA), an active metabolite of dipyrone. Twelve healthy adult male volunteers received single 600 mg oral doses of dipyrone in two formulations at a 7-day interval in a randomized, crossover protocol. Plasma concentrations of MAA (N = 336), measured by HPLC, were used to develop LSS models. Linear regression analysis and a "jack-knife" validation procedure revealed that the AUC0-¥ and the Cmax of MAA can be accurately predicted (R²>0.95, bias <1.5%, precision between 3.1 and 8.3%) by LSS models based on two sampling times. Validation tests indicate that the most informative 2-point LSS models developed for one formulation provide good estimates (R²>0.85) of the AUC0-¥ or Cmax for the other formulation. LSS models based on three sampling points (1.5, 4 and 24 h), but using different coefficients for AUC0-¥ and Cmax, predicted the individual values of both parameters for the enrolled volunteers (R²>0.88, bias = -0.65 and -0.37%, precision = 4.3 and 7.4%) as well as for plasma concentration data sets generated by simulation (R²>0.88, bias = -1.9 and 8.5%, precision = 5.2 and 8.7%). Bioequivalence assessment of the dipyrone formulations based on the 90% confidence interval of log-transformed AUC0-¥ and Cmax provided similar results when either the best-estimated or the LSS-derived metrics were used.

Environmental formaldehyde analysis by active diffusive sampling with a bundle of polypropylene porous capillaries followed by capillary zone electrophoretic separation and contactless conductivity detection

Compared to other volatile carbonylic compounds present in outdoor air, formaldehyde (CH2O) is the most toxic, deserving more attention in terms of indoor and outdoor air quality legislation and control. The analytical determination of CH2O in air still presents challenges due to the low-level concentration (in the sub-ppb range) and its variation with sampling site and time. Of the many available analytical methods for carbonylic compounds, the most widespread one is the time consuming collection in cartridges impregnated with 2,4-dinitrophenylhydrazine followed by the analysis of the formed hydrazones by HPLC. The present work proposes the use of polypropylene hollow porous capillary fibers to achieve efficient CH2O collection. The Oxyphan (R) fiber (designed for blood oxygenation) was chosen for this purpose because it presents good mechanical resistance, high density of very fine pores and high ratio of collection area to volume of the acceptor fluid in the tube, all favorable for the development of air sampling apparatus. The collector device consists of a Teflon pipe inside of which a bundle of polypropylene microporous capillary membranes was introduced. While the acceptor passes at a low flow rate through the capillaries, the sampled air circulates around the fibers, impelled by a low flow membrane pump (of the type used for aquariums ventilation). The coupling of this sampling technique with the selective and quantitative determination of CH2O, in the form of hydroxymethanesulfonate (HMS) after derivatization with HSO3-, by capillary electrophoresis with capacitively coupled contactless conductivity detection (CE-(CD)-D-4) enabled the development of a complete analytical protocol for the CH2O evaluation in air. (C) 2008 Published by Elsevier B.V.

Prevalence of trachoma in a population of the upper Rio Negro Basin and risk factors for active disease

Purpose: To determine the prevalence of trachoma in Sao Gabriel da Cachoeira (SGC), the only urban community of the upper Rio Negro Basin of the Amazon state in Brazil, near the Colombian border, and to investigate the risk factors associated with the active forms of the disease. Methods: A total of 1702 people (440 children up to 9 years and 1069 adults aged 15 years and above) were examined. The sample was selected from a probabilistic household sampling procedure based on census data and a previous study of trachoma prevalence in Sao Gabriel da Cachoeira. A two-stage probabilistic household cluster sample was drawn. Household units were randomly selected within each cluster. A variety of socioeconomic and hygiene variables were studied in order to determine the risk factors for active trachoma in a household. Results: The total prevalence of trachoma was 8.9%. Prevalence of active trachoma (TF and/or TI) in children aged 1-9 years was 11.1% and trachomatous trichiasis in adults aged 15 years and above was 0.19%. Trachomatous scarring reached a peak of 22.4% for subjects between 50 to 60 years of age. Corneal opacity occurred in subjects aged 50 years and older with a prevalence of 2.0%. No sex effect was found on the overall prevalence of trachoma in SGC. Risk factors associated with active trachoma were mainly related to poor socioeconomic indicators. Conclusions: Despite the ubiquitous presence of water, the analysis of the risk factors associated with the active forms of the disease supports the idea that a low personal standard of hygiene and not water availability per se, is the key factor associated with trachoma.

Implementation and comparison of different switching techniques for shunt active power filters

This paper presents a comparison between three switching techniques that can be used in three-phase four-wire Shunt Active Power Filters (SAPFs). The implemented switching techniques are: Periodic-Sampling (PS), Triangular Carrier Pulse-Width Modulation (TC-PWM) and Space Vector PWM (SVPWM). The comparison between them is made in terms of the compensated currents THD%, implementation complexity, necessary CPU time and SAPF efficiency. To perform this comparison are presented and analyzed several experimental results, obtained with a 20 kVA Shunt Active Power Filter prototype, specially developed for this purpose. The control system of the developed SAPF is based in the p-q Theory with a grid synchronization algorithm p-PLL.

Experimental validation of a proposed single-phase five-level active rectifier operating with model predictive current control

This paper presents a model predictive current control applied to a proposed single-phase five-level active rectifier (FLAR). This current control strategy uses the discrete-time nature of the active rectifier to define its state in each sampling interval. Although the switching frequency is not constant, this current control strategy allows to follow the reference with low total harmonic distortion (THDF). The implementation of the active rectifier that was used to obtain the experimental results is described in detail along the paper, presenting the circuit topology, the principle of operation, the power theory, and the current control strategy. The experimental results confirm the robustness and good performance (with low current THDF and controlled output voltage) of the proposed single-phase FLAR operating with model predictive current control.

A Survey of Active Learning Algorithms for Supervised Remote Sensing Image Classification

Defining an efficient training set is one of the most delicate phases for the success of remote sensing image classification routines. The complexity of the problem, the limited temporal and financial resources, as well as the high intraclass variance can make an algorithm fail if it is trained with a suboptimal dataset. Active learning aims at building efficient training sets by iteratively improving the model performance through sampling. A user-defined heuristic ranks the unlabeled pixels according to a function of the uncertainty of their class membership and then the user is asked to provide labels for the most uncertain pixels. This paper reviews and tests the main families of active learning algorithms: committee, large margin, and posterior probability-based. For each of them, the most recent advances in the remote sensing community are discussed and some heuristics are detailed and tested. Several challenging remote sensing scenarios are considered, including very high spatial resolution and hyperspectral image classification. Finally, guidelines for choosing the good architecture are provided for new and/or unexperienced user.

Antinucleosome antibodies as a marker of active proliferative lupus nephritis.

BACKGROUND: Antinucleosome autoantibodies were previously described to be a marker of active lupus nephritis. However, the true prevalence of antinucleosome antibodies at the time of active proliferative lupus nephritis has not been well established. Therefore, the aim of this study is to define the prevalence and diagnostic value of autoantibodies against nucleosomes as a marker for active proliferative lupus nephritis. STUDY DESIGN: Prospective multicenter diagnostic test study. SETTING & PARTICIPANTS: 35 adult patients with systemic lupus erythematosus (SLE) at the time of the renal biopsy showing active class III or IV lupus nephritis compared with 59 control patients with SLE. INDEX TEST: Levels of antinucleosome antibodies and anti-double-stranded DNA (anti-dsDNA) antibodies. REFERENCE TEST: Kidney biopsy findings of class III or IV lupus nephritis at the time of sampling in a study population versus clinically inactive or no nephritis in a control population. RESULTS: Increased concentrations of antinucleosome antibodies were found in 31 of 35 patients (89%) with active proliferative lupus nephritis compared with 47 of 59 control patients (80%) with SLE. No significant difference between the 2 groups with regard to number of positive patients (P = 0.2) or antibody concentrations (P = 0.2) could be found. The area under the receiver operating characteristic curve as a marker of the accuracy of the test in discriminating between proliferative lupus nephritis and inactive/no nephritis in patients with SLE was 0.581 (95% confidence interval, 0.47 to 0.70; P = 0.2). Increased concentrations of anti-dsDNA antibodies were found in 33 of 35 patients (94.3%) with active proliferative lupus nephritis compared with 49 of 58 control patients (84.5%) with SLE (P = 0.2). In patients with proliferative lupus nephritis, significantly higher titers of anti-dsDNA antibodies were detected compared with control patients with SLE (P < 0.001). The area under the receiver operating characteristic curve in discriminating between proliferative lupus nephritis and inactive/no nephritis in patients with SLE was 0.710 (95% confidence interval, 0.60 to 0.82; P < 0.001). CONCLUSIONS: Antinucleosome antibodies have a high prevalence in patients with severe lupus nephritis. However, our data suggest that determining antinucleosome antibodies is of limited help in the distinction of patients with active proliferative lupus nephritis from patients with SLE without active renal disease.

When rare is just a matter of sampling: unexpected dominance of clubtail dragonflies (Odonata, Gomphidae) through different collecting methods at Parque Nacional da Serra do Cipó, Minas Gerais State, Brazil

When rare is just a matter of sampling: Unexpected dominance of clubtail dragonflies (Odonata, Gomphidae) through different collecting methods at Parque Nacional da Serra do Cipó, Minas Gerais State, Brazil. Capture of dragonfly adults during two short expeditions to Parque Nacional da Serra do Cipó, Minas Gerais State, using three distinct collecting methodsaerial nets, Malaise and light sheet trapsis reported. The results are outstanding due the high number of species of Gomphidae (7 out of 26 Odonata species), including a new species of Cyanogomphus Selys, 1873, obtained by two non-traditional collecting methods. Because active collecting with aerial nets is the standard approach for dragonfly inventories, we discuss some aspects of the use of traps, comparing our results with those in the literature, suggesting they should be used as complementary methods in faunistic studies. Furthermore, Zonophora campanulata annulata Belle, 1983 is recorded for the first time from Minas Gerais State and taxonomic notes about Phyllogomphoides regularis (Selys, 1873) and Progomphus complicatus Selys, 1854 are also given.

Active learning methods for remote sensing image classification

In this paper, we propose two active learning algorithms for semiautomatic definition of training samples in remote sensing image classification. Based on predefined heuristics, the classifier ranks the unlabeled pixels and automatically chooses those that are considered the most valuable for its improvement. Once the pixels have been selected, the analyst labels them manually and the process is iterated. Starting with a small and nonoptimal training set, the model itself builds the optimal set of samples which minimizes the classification error. We have applied the proposed algorithms to a variety of remote sensing data, including very high resolution and hyperspectral images, using support vector machines. Experimental results confirm the consistency of the methods. The required number of training samples can be reduced to 10% using the methods proposed, reaching the same level of accuracy as larger data sets. A comparison with a state-of-the-art active learning method, margin sampling, is provided, highlighting advantages of the methods proposed. The effect of spatial resolution and separability of the classes on the quality of the selection of pixels is also discussed.

EADock: docking of small molecules into protein active sites with a multiobjective evolutionary optimization.

In recent years, protein-ligand docking has become a powerful tool for drug development. Although several approaches suitable for high throughput screening are available, there is a need for methods able to identify binding modes with high accuracy. This accuracy is essential to reliably compute the binding free energy of the ligand. Such methods are needed when the binding mode of lead compounds is not determined experimentally but is needed for structure-based lead optimization. We present here a new docking software, called EADock, that aims at this goal. It uses an hybrid evolutionary algorithm with two fitness functions, in combination with a sophisticated management of the diversity. EADock is interfaced with the CHARMM package for energy calculations and coordinate handling. A validation was carried out on 37 crystallized protein-ligand complexes featuring 11 different proteins. The search space was defined as a sphere of 15 A around the center of mass of the ligand position in the crystal structure, and on the contrary to other benchmarks, our algorithm was fed with optimized ligand positions up to 10 A root mean square deviation (RMSD) from the crystal structure, excluding the latter. This validation illustrates the efficiency of our sampling strategy, as correct binding modes, defined by a RMSD to the crystal structure lower than 2 A, were identified and ranked first for 68% of the complexes. The success rate increases to 78% when considering the five best ranked clusters, and 92% when all clusters present in the last generation are taken into account. Most failures could be explained by the presence of crystal contacts in the experimental structure. Finally, the ability of EADock to accurately predict binding modes on a real application was illustrated by the successful docking of the RGD cyclic pentapeptide on the alphaVbeta3 integrin, starting far away from the binding pocket.

Evaluation of potential breath biomarkers for active smoking: assessment of smoking habits

Different compounds have been reported as biomarkers of a smoking habit, but, to date, there is no appropriate biomarker for tobacco-related exposure because the proposed chemicals seem to be nonspecific or they are only appropriate for short-term exposure. Moreover, conventional sampling methodologies require an invasive method because blood or urine samples are required. The use of a microtrap system coupled to gas chromatography–mass spectrometry analysis has been found to be very effective for the noninvasive analysis of volatile organic compounds in breath samples. The levels of benzene, 2,5-dimethylfuran, toluene, o-xylene, and m- p-xylene have been analyzed in breath samples obtained from 204 volunteers (100 smokers, 104 nonsmokers; 147 females, 57 males; ages 16 to 53 years). 2,5-Dimethylfuran was always below the limit of detection (0.005 ppbv) in the nonsmoker population and always detected in smokers independently of the smoking habits. Benzene was only an effective biomarker for medium and heavy smokers, and its level was affected by smoking habits. Regarding the levels of xylenes and toluene, they were only different in heavy smokers and after short-term exposure. The results obtained suggest that 2,5-dimethylfuran is a specific breath biomarker of smoking status independently of the smoking habits (e.g., short- and long-term exposure, light and heavy consumption), and so this compound might be useful as a biomarker of smoking exposure

Evaluation of potential breath biomarkers for active smoking: assessment of smoking habits

Different compounds have been reported as biomarkers of a smoking habit, but, to date, there is no appropriate biomarker for tobacco-related exposure because the proposed chemicals seem to be nonspecific or they are only appropriate for short-term exposure. Moreover, conventional sampling methodologies require an invasive method because blood or urine samples are required. The use of a microtrap system coupled to gas chromatography–mass spectrometry analysis has been found to be very effective for the noninvasive analysis of volatile organic compounds in breath samples. The levels of benzene, 2,5-dimethylfuran, toluene, o-xylene, and m- p-xylene have been analyzed in breath samples obtained from 204 volunteers (100 smokers, 104 nonsmokers; 147 females, 57 males; ages 16 to 53 years). 2,5-Dimethylfuran was always below the limit of detection (0.005 ppbv) in the nonsmoker population and always detected in smokers independently of the smoking habits. Benzene was only an effective biomarker for medium and heavy smokers, and its level was affected by smoking habits. Regarding the levels of xylenes and toluene, they were only different in heavy smokers and after short-term exposure. The results obtained suggest that 2,5-dimethylfuran is a specific breath biomarker of smoking status independently of the smoking habits (e.g., short- and long-term exposure, light and heavy consumption), and so this compound might be useful as a biomarker of smoking exposure

Variational Monte Carlo estimation of the dissociation energy of CuH using correlated sampling

A new approach to treating large Z systems by quantum Monte Carlo has been developed. It naturally leads to notion of the 'valence energy'. Possibilities of the new approach has been explored by optimizing the wave function for CuH and Cu and computing dissociation energy and dipole moment of CuH using variational Monte Carlo. The dissociation energy obtained is about 40% smaller than the experimental value; the method is comparable with SCF and simple pseudopotential calculations. The dipole moment differs from the best theoretical estimate by about 50% what is again comparable with other methods (Complete Active Space SCF and pseudopotential methods).

Sequential Optimal Recovery: A Paradigm for Active Learning

In most classical frameworks for learning from examples, it is assumed that examples are randomly drawn and presented to the learner. In this paper, we consider the possibility of a more active learner who is allowed to choose his/her own examples. Our investigations are carried out in a function approximation setting. In particular, using arguments from optimal recovery (Micchelli and Rivlin, 1976), we develop an adaptive sampling strategy (equivalent to adaptive approximation) for arbitrary approximation schemes. We provide a general formulation of the problem and show how it can be regarded as sequential optimal recovery. We demonstrate the application of this general formulation to two special cases of functions on the real line 1) monotonically increasing functions and 2) functions with bounded derivative. An extensive investigation of the sample complexity of approximating these functions is conducted yielding both theoretical and empirical results on test functions. Our theoretical results (stated insPAC-style), along with the simulations demonstrate the superiority of our active scheme over both passive learning as well as classical optimal recovery. The analysis of active function approximation is conducted in a worst-case setting, in contrast with other Bayesian paradigms obtained from optimal design (Mackay, 1992).