Conventional Ensemble Methods Research Articles

Temperature Cycles Unravel the Dynamics of Single Biomolecules

Temperature Cycles Unravel the Dynamics of Single Biomolecules

Convex ensemble learning with sparsity and diversity

Classifier ensemble has been broadly studied in two prevalent directions, i.e., to diversely generate classifier components, and to sparsely combine multiple classifiers. While most current approaches are emphasized on either sparsity or diversity only, we investigate classifier ensemble focused on both in this paper. We formulate the classifier ensemble problem with the sparsity and diversity learning in a general mathematical framework, which proves beneficial for grouping classifiers. In particular, derived from the error-ambiguity decomposition, we design a convex ensemble diversity measure. Consequently, accuracy loss, sparseness regularization, and diversity measure can be balanced and combined in a convex quadratic programming problem. We prove that the final convex optimization leads to a closed-form solution, making it very appealing for real ensemble learning problems. We compare our proposed novel method with other conventional ensemble methods such as Bagging, least squares combination, sparsity learning, and AdaBoost, extensively on a variety of UCI benchmark data sets and the Pascal Large Scale Learning Challenge 2008 webspam data. Experimental results confirm that our approach has very promising performance.

Single Lipid Vesicle Assay for Characterizing Single-Enzyme Kinetics of Phospholipid Hydrolysis in a Complex Biological Fluid

Imaging of individual lipid vesicles is used to track single-enzyme kinetics of phospholipid hydrolysis. The method is employed to quantify the catalytic activity of phospholipase A2 (PLA2) in both pure and complex biological fluids. The measurements are demonstrated to offer a subpicomolar limit of detection (LOD) of human secretory PLA2 (sPLA2) in up to 1000-fold-diluted cerebrospinal fluid (CSF). An additional new feature provided by the single-enzyme sensitivity is that information about both relative concentration variations of active sPLA2 in CSF and the specific enzymatic activity can be simultaneously obtained. When CSF samples from healthy controls and individuals diagnosed with Alzheimer's disease (AD) are analyzed, the specific enzymatic activity is found to be preserved within 7% in the different CSF samples whereas the enzyme concentration differs by up to 56%. This suggests that the previously reported difference in PLA2 activity in CSF samples from healthy and AD individuals originates from differences in the PLA2 expression level rather than from the enzyme activity. Conventional ensemble averaging methods used to probe sPLA2 activity do not allow one to obtain such information. Together with an improvement in the LOD of at least 1 order of magnitude compared to that of conventional assays, this suggests that the method will become useful in furthering our understanding of the role of PLA2 in health and disease and in detecting the pharmacodynamic effects of PLA2-targeting drug candidates.

An efficient ensemble learning method for gene microarray classification.

The gene microarray analysis and classification have demonstrated an effective way for the effective diagnosis of diseases and cancers. However, it has been also revealed that the basic classification techniques have intrinsic drawbacks in achieving accurate gene classification and cancer diagnosis. On the other hand, classifier ensembles have received increasing attention in various applications. Here, we address the gene classification issue using RotBoost ensemble methodology. This method is a combination of Rotation Forest and AdaBoost techniques which in turn preserve both desirable features of an ensemble architecture, that is, accuracy and diversity. To select a concise subset of informative genes, 5 different feature selection algorithms are considered. To assess the efficiency of the RotBoost, other nonensemble/ensemble techniques including Decision Trees, Support Vector Machines, Rotation Forest, AdaBoost, and Bagging are also deployed. Experimental results have revealed that the combination of the fast correlation-based feature selection method with ICA-based RotBoost ensemble is highly effective for gene classification. In fact, the proposed method can create ensemble classifiers which outperform not only the classifiers produced by the conventional machine learning but also the classifiers generated by two widely used conventional ensemble learning methods, that is, Bagging and AdaBoost.

Time-expanded sampling for ensemble-based filters: Assimilation experiments with real radar observations

By sampling perturbed state vectors from each ensemble prediction run at properly selected time levels in the vicinity of the analysis time, the recently proposed time-expanded sampling approach can enlarge the ensemble size without increasing the number of prediction runs and, hence, can reduce the computational cost of an ensemble-based filter. In this study, this approach is tested for the first time with real radar data from a tornadic thunderstorm. In particular, four assimilation experiments were performed to test the time-expanded sampling method against the conventional ensemble sampling method used by ensemble-based filters. In these experiments, the ensemble square-root filter (EnSRF) was used with 45 ensemble members generated by the time-expanded sampling and conventional sampling from 15 and 45 prediction runs, respectively, and quality-controlled radar data were compressed into super-observations with properly reduced spatial resolutions to improve the EnSRF performances. The results show that the time-expanded sampling approach not only can reduce the computational cost but also can improve the accuracy of the analysis, especially when the ensemble size is severely limited due to computational constraints for real-radar data assimilation. These potential merits are consistent with those previously demonstrated by assimilation experiments with simulated data.

Active Diverse Learning Neural Network Ensemble Approach for Power Transformer Fault Diagnosis

A n ensemble learning algorithm was proposed in this paper by analyzing the error function of neural network ensembles, by which , individual neural networks were actively guided to learn divers ity . By decomposing the ensemble error function, error correlation terms were included in the learning criterion function of individual networks. And all the individual networks in the ensemble were leaded to learn divers ity through cooperative training. The method was applied in Dissolved Gas Analysis based fault diagnosis of power transformer. Experiment results show that, the algorithm has higher accuracy than IEC method and BP network. In addition, the performance is more stable than conventional ensemble method, i.e., Bagging and Boosting.

Single-molecule, real-time measurement of enzyme kinetics by alternating-laser excitation fluorescence resonance energy transfer

Using a single-molecule fluorescence method uniquely suitable for binding assay, alternating-laser excitation fluorescence resonance energy transfer (ALEX-FRET), we accurately measured the cleavage rate of 8-17 deoxyribozyme, an RNA-cleaving enzyme, at the single-molecule level in real time with a minimum consumption of samples, i.e., at least three orders of magnitude smaller than used in the conventional ensemble FRET method.

A novel ensemble machine learning for robust microarray data classification

Microarray data analysis and classification has demonstrated convincingly that it provides an effective methodology for the effective diagnosis of diseases and cancers. Although much research has been performed on applying machine learning techniques for microarray data classification during the past years, it has been shown that conventional machine learning techniques have intrinsic drawbacks in achieving accurate and robust classifications. This paper presents a novel ensemble machine learning approach for the development of robust microarray data classification. Different from the conventional ensemble learning techniques, the approach presented begins with generating a pool of candidate base classifiers based on the gene sub-sampling and then the selection of a sub-set of appropriate base classifiers to construct the classification committee based on classifier clustering. Experimental results have demonstrated that the classifiers constructed by the proposed method outperforms not only the classifiers generated by the conventional machine learning but also the classifiers generated by two widely used conventional ensemble learning methods (bagging and boosting).

Tracking of single molecules as a powerful method to characterize diffusivity of organic species in mesoporous materials

The diffusion of individual fluorescent molecules can be observed by single-molecule tracking techniques and characterized by the analysis of their diffusional trajectory. Heterogeneities in the diffusivity that would pass undetected by conventional ensemble methods or fluorescence correlation spectroscopy are resolved by this method. This is demonstrated using four different examples in which we analyse the diffusion of single organic dye molecules in mesoporous materials. We show that this method can be used to obtain structural information from the inner structure of nanoporous materials with a resolution better than the optical diffraction limit.

Comparative quantification of nucleic acids using single-molecule detection and molecular beacons

This paper reports a highly sensitive homogenous method for comparative quantification of nucleic acids based on single-molecule detection (SMD) and molecular beacons (MBs). Two different color MBs were used to perform a separation-free comparative hybridization assay for simultaneous quantification of both target and control strands. A fluorescent burst, emitted from a single hybrid when it passes through a minuscule laser-focused region, is detected with high signal-to-noise ratio (SNR) by using single-molecule fluorescence spectroscopy. Targets are quantified via counting of discrete fluorescent bursts. The high SNR achieved in both detection channels overcame the complications of fluorescent variability usually observed in dual-color ensemble measurements. In comparison with the conventional ensemble methods, this method improved the detection limit by 3 orders of magnitude and reduced the probe consumption by 6 orders of magnitude, facilitating a highly sensitive approach for comparative quantification of nucleic acids and offering great promise for genomic quantification without amplification.

Monte Carlo Simulation Study of Adsorption Characteristics in Slit-Like Micropores Under Supercritical Conditions

Abstract Adsorption characteristics of a solute diluted in supercritical fluids has been investigated by using the Monte Carlo simulation techniques. The Lennard-Jones potential function is used for describing interactions for a model system of CO2 + benzene in slit-like micropores with infinite graphitic carbon walls. A modified μVT ensemble method with particle exchange proposed by Cracknell, Nicholson and Quirke (1993) is found to be much superior to the conventional μVT ensemble method especially for dense mixtures in a pore. Adsorption isotherms of CO2 and benzene, in equilibrium with a dilute benzene mixture in CO2 (mole fraction of benzene = 0.001), are computed by varying pressure, temperature, the benzene–surface interaction potential, and the slitwidth. Adsorption isotherm curve of CO2 increases with an increase in pressure while that of benzene shows a maximum at a pressure far below the critical pressure of CO2 and then it decreases with increasing pressure. The decrease in benzene adsorption with increasing pressure is attributable to both the enhanced solubility in supercritical CO2 and the competitive adsorption of CO2. The isotherm curves of each component at two temperatures, 313.2 K and 323.2 K, show to cross at a pressure near the critical pressure due to the “density effect” on the chemical potentials of a solute at supercritical fluid conditions. When the interaction between a solute and a surface increases, the adsorption isotherm increases. Narrowing the slitwidth results in the increase in the adsorption of solute since the external potential from two walls becomes deeper.