Search

One of the main challenges in cancer surgery is ensuring that all tumour cells are removed during surgery, while sparing as much healthy tissue as possible. Histopathology, the gold-standard technique for cancer diagnosis, is often impractical for intra-operative use because of the time-consuming tissue preparation procedures (sectioning and staining). Raman micro-spectroscopy is a powerful technique that can discriminate between tumours and healthy tissues with high accuracy, based entirely on intrinsic chemical differences. However, raster-scanning Raman micro-spectroscopy is a slow imaging technique that typically requires data acquisition times as long as several days for typical tissue samples obtained during surgery (1 x 1 cm²) - in particular when high signal-to-noise ratio spectra are required to ensure accurate diagnosis. In this paper we present two techniques based on selective sampling Raman micro-spectroscopy that can overcome these limitations. In selective sampling, information regarding the spatial features of the tissue, either measured by an alternative optical technique or estimated in real-time from the Raman spectra, can be used to drastically reduce the number of Raman spectra required for diagnosis. These sampling strategies allowed diagnosis of basal cell carcinoma in skin tissue samples excised during Mohs micrographic surgery faster than frozen section histopathology, and two orders of magnitude faster than previous techniques based on raster-scanning Raman microscopy. Further development of these techniques may help during cancer surgery by providing a fast and objective way for surgeons to ensure the complete removal of tumour cells while sparing as much healthy tissue as possible. [ABSTRACT FROM AUTHOR]

Abstract: Relevance feedback is an effective approach to improve the performance of image retrieval by leveraging the labeling of human. In order to alleviate the burden of labeling, active learning method has been introduced to select the most informative samples for labeling. In this paper, we present a novel batch mode active learning scheme for informative sample selection. Inspired by the method of graph propagation, we not only take the correlation between labeled samples and unlabeled samples, but the correlation among unlabeled samples taken into account as well. Especially, considering the unbalanced distribution of samples and the personalized feedback of human we propose an asymmetric propagation scheme to unify the various criteria including uncertainty, diversity and density into batch mode active learning in relevance feedback. Extensive experiments on publicly available datasets show that the proposed method is promising. [Copyright &y& Elsevier]

A method for selective sampling and analysis of explosive residues on solid surfaces based on laser-induced breakdown spectroscopy (LIBS) is presented. Organic explosives are difficult to analyze when present as residues on organic materials. Under these circumstances LIBS suffers from the limitations imposed by the limited spectroscopic information available for the analysis. Since ablation and subsequent plasma formation are sensitive to the beam focal conditions and the pulse energy deposited on the surface, the choice of an appropriate set of experimental conditions increases the surface sensitivity of the analysis and hence a selective inspection of the residue in the absence of spectral contribution from the organic support analyzed. 2-Mononitrotoluene (MNT), 2,6- dinitrotoluene (DNT), and 2,4,6-trinitrotoluene (TNT) are used as model residues, whereas nylon and Teflon are used as illustrative surfaces of daily life objects. The results demonstrate that selective sampling is successfully achieved in all cases when the plasma formation threshold of the residues and the object is substantially different. Plasma imaging demonstrates that the species distribution along the plume changes with beam focal conditions, which is exploited here to further increase the selectivity of the approach. [ABSTRACT FROM AUTHOR]

The current literature on so-called 'skew-symmetric distributions' is closely linked to the idea of a selection mechanism operated by some latent variable. We illustrate the pioneering work of Fernando de Helguero who in 1908 put forward a formulation for the genesis of non-normal distributions via a selection mechanism, which perturbs a normal distribution, hence employing a closely connected argument with the one now widely used in this context. Arguably, de Helguero can then be considered the precursor of the current idea of skew-symmetric distributions. Unfortunately, a tragic quirk of fate did not allow him to pursue his project beyond the initial formulation and his work went unnoticed for the rest of the 20th century. [ABSTRACT FROM AUTHOR]

In this paper, we propose a method that can identify challenging music samples for beat tracking without ground truth. Our method, motivated by the machine learning method “selective sampling,” is based on the measurement of mutual agreement between beat sequences. In calculating this mutual agreement we show the critical influence of different evaluation measures. Using our approach we demonstrate how to compile a new evaluation dataset comprised of difficult excerpts for beat tracking and examine this difficulty in the context of perceptual and musical properties. Based on tag analysis we indicate the musical properties where future advances in beat tracking research would be most profitable and where beat tracking is too difficult to be attempted. Finally, we demonstrate how our mutual agreement method can be used to improve beat tracking accuracy on large music collections. [ABSTRACT FROM PUBLISHER]

We study the theoretical advantages of active learning over passive learning. Specifically, we prove that, in noise-free classifier learning for VC classes, any passive learning algorithm can be transformed into an active learning algorithm with asymptotically strictly superior label complexity for all nontrivial target functions and distributions. We further provide a general characterization of the magnitudes of these improvements in terms of a novel generalization of the disagreement co-efficient. We also extend these results to active learning in the presence of label noise, and find that even under broad classes of noise distributions, we can typically guarantee strict improvements over the known results for passive learning. [ABSTRACT FROM AUTHOR]

We discover a strong relation between two known learning models: stream-based active learning and perfect selective classification (an extreme case of 'classification with a reject option'). For these models, restricted to the realizable case, we show a reduction of active learning to selective classification that preserves fast rates. Applying this reduction to recent results for selective classification, we derive exponential target-independent label complexity speedup for actively learning general (non-homogeneous) linear classifiers when the data distribution is an arbitrary high dimensional mixture of Gaussians. Finally, we study the relation between the proposed technique and existing label complexity measures, including teaching dimension and disagreement coefficient. [ABSTRACT FROM AUTHOR]

We present a new active learning algorithm based on nonparametric estimators of the regression function. Our investigation provides probabilistic bounds for the rates of convergence of the generalization error achievable by proposed method over a broad class of underlying distributions. We also prove minimax lower bounds which show that the obtained rates are almost tight. [ABSTRACT FROM AUTHOR]

Objective: Genome-wide association studies have identified robust associations between single nucleotide polymorphisms and complex traits. As the proportion of phenotypic variance explained is still limited for most of the traits, larger and larger meta-analyses are being conducted to detect additional associations. Here we investigate the impact of the study design and the underlying assumption about the true genetic effect in a bimodal mixture situation on the power to detect associations. Methods: We performed simulations of quantitative phenotypes analysed by standard linear regression and dichotomized case-control data sets from the extremes of the quantitative trait analysed by standard logistic regression. Results: Using linear regression, markers with an effect in the extremes of the traits were almost undetectable, whereas analysing extremes by case-control design had superior power even for much smaller sample sizes. Two real data examples are provided to support our theoretical findings and to explore our mixture and parameter assumption. Conclusions: Our findings support the idea to re-analyse the available meta-analysis data sets to detect new loci in the extremes. Moreover, our investigation offers an explanation for discrepant findings when analysing quantitative traits in the general population and in the extremes. Copyright © 2011 S. Karger AG, Basel [ABSTRACT FROM AUTHOR]

We introduce efficient margin-based algorithms for selective sampling and filtering in binary classification tasks. Experiments on real-world textual data reveal that our algorithms perform significantly better than popular and similarly efficient competitors. Using the so-called Mammen-Tsybakov low noise condition to parametrize the instance distribution, and assuming linear label noise, we show bounds on the convergence rate to the Bayes risk of a weaker adaptive variant of our selective sampler. Our analysis reveals that, excluding logarithmic factors, the average risk of this adaptive sampler converges to the Bayes risk at rate N where N denotes the number of queried labels, and α>0 is the exponent in the low noise condition. For all $\alpha>\sqrt{3}-1\approx0.73$ this convergence rate is asymptotically faster than the rate N achieved by the fully supervised version of the base selective sampler, which queries all labels. Moreover, for α→∞ (hard margin condition) the gap between the semi- and fully-supervised rates becomes exponential. [ABSTRACT FROM AUTHOR]

Abstract: Pairwise clustering methods have shown great promise for many real-world applications. However, the computational demands of these methods make them impractical for use with large data sets. The contribution of this paper is a simple but efficient method, called eSPEC, that makes clustering feasible for problems involving large data sets. Our solution adopts a “sampling, clustering plus extension” strategy. The methodology starts by selecting a small number of representative samples from the relational pairwise data using a selective sampling scheme; then the chosen samples are grouped using a pairwise clustering algorithm combined with local scaling; and finally, the label assignments of the remaining instances in the data are extended as a classification problem in a low-dimensional space, which is explicitly learned from the labeled samples using a cluster-preserving graph embedding technique. Extensive experimental results on several synthetic and real-world data sets demonstrate both the feasibility of approximately clustering large data sets and acceleration of clustering in loadable data sets of our method. [ABSTRACT FROM AUTHOR]

many databases have been brought online, data retrieval-finding relevant data from large databases-has become a nontrivial task. A feedback-based data retrieval system was proposed to provide user with an intuitive way for expressing their preferences in queries. The system iteratively receives a partial ordering on a sample of data from the user, learns a ranking function, and returns highly ranked results according to the function. An important issue in such retrieval systems is minimizing the number of iterations or the amount of feedback to learn an accurate ranking function. This paper proposes selective sampling (or active learning) techniques for RankSVM that can be used in the retrieval systems. The proposed techniques minimizes the amount of user interaction to learn an accurate ranking function thus facilitates users formulating a preference query in the data retrieval system. [ABSTRACT FROM AUTHOR]

We describe and explore a new perspective on the sample complexity of active learning. In many situations where it was generally believed that active learning does not help, we show that active learning does help in the limit, often with exponential improvements in sample complexity. This contrasts with the traditional analysis of active learning problems such as non-homogeneous linear separators or depth-limited decision trees, in which Ω(1/ ε) lower bounds are common. Such lower bounds should be interpreted carefully; indeed, we prove that it is always possible to learn an ε-good classifier with a number of samples asymptotically smaller than this. These new insights arise from a subtle variation on the traditional definition of sample complexity, not previously recognized in the active learning literature. [ABSTRACT FROM AUTHOR]

This study investigates the potential and limitations of a field-deployable analytical approach that involves selective capture of explosive materials with direct analysis by ion mobility spectrometry (IMS). Selective capture of explosives was performed on deactivated quartz fiber filters impregnated with metal β-diketonate polymers. These Lewis acidic polymers selectively interact with Lewis base analytes such as explosives. The filters were directly inserted into an IMS instrument for analysis. The uptake kinetics of 2,4,6-trinitrotoluene (TNT) from a saturated atmosphere were characterized, and based on these studies, passive equilibrium sampling was applied to estimate the TNT concentration within an ammunition magazine that contained bulk TNT. Hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) uptake from a saturated environment also was examined over a one-month period. Each incremental sampling period showed increasing quantities of RDX culminating with collection of approximately 5 ng of RDX on the filter at the end of one month. This is the first time that gas-phase uptake of RDX has been demonstrated. [ABSTRACT FROM AUTHOR]

Abstract: A novel approach for sampling k -space in a pure phase encoding imaging sequence is presented using the Single Point Imaging (SPI) technique. The sequence is optimised with respect to the achievable Signal-to-Noise ratio (SNR) for a given time interval via selective sparse k -space sampling, dictated by prior knowledge of the overall object of interest’s shape. This allows dynamic processes featuring short NMR signal to be more readily followed, in our case the absorption of moisture by a cereal-based wafer material. Further improvements in image quality are also shown via the use of complete sampling of k -space at the start or end of the series of imaging experiments; followed by subsequent use of this data for un-sampled k -space points as opposed to zero filling. [Copyright &y& Elsevier]

We survey a number of recent results concerning the behaviour of algorithms for learning classifiers based on the solution of a regularized least-squares problem. [Copyright &y& Elsevier]

Abstract: In relevance feedback algorithms, selective sampling is often used to reduce the cost of labeling and explore the unlabeled data. In this paper, we proposed an active learning algorithm, Co-SVM, to improve the performance of selective sampling in image retrieval. In Co-SVM algorithm, color and texture are naturally considered as sufficient and uncorrelated views of an image. SVM classifiers are learned in color and texture feature subspaces, respectively. Then the two classifiers are used to classify the unlabeled data. These unlabeled samples which are differently classified by the two classifiers are chose to label. The experimental results show that the proposed algorithm is beneficial to image retrieval. [Copyright &y& Elsevier]

A selective sampling algorithm is a learning algorithm for classification that, based on the past observed data, decides whether to ask the label of each new instance to be classified. In this paper, we introduce a general technique for turning linear-threshold classification algorithms from the general additive family into randomized selective sampling algorithms. For the most popular algorithms in this family we derive mistake bounds that hold for individual sequences of examples. These bounds show that our semi-supervised algorithms can achieve, on average, the same accuracy as that of their fully supervised counterparts, but using fewer labels. Our theoretical results are corroborated by a number of experiments on real-world textual data. The outcome of these experiments is essentially predicted by our theoretical results: Our selective sampling algorithms tend to perform as well as the algorithms receiving the true label after each classification, while observing in practice substantially fewer labels. [ABSTRACT FROM AUTHOR]

Abstract: This paper presents adaptive boosting with uncertainty-based selective sampling (AdaBUS), a variant of the AdaBoost algorithm for boosting the Naı¨ve Bayes (NB) text classification. Although the boosting technique has been shown to effectively improve the accuracy of machine-learning-based classifiers, boosting does not work well with NB text classification owing to the low variance in the accuracy of its base classifier. In this study, we propose boosting the NB text classifier by combining the AdaBoost boosting algorithm with uncertainty-based selective sampling. Experiments using the popular Reuters-21578 document collection showed that the proposed algorithm effectively improves classification accuracy. [Copyright &y& Elsevier]

This paper provides an introductory overview of a portion of distribution theory which is currently under intense development. The starting point of this topic has been the so-called skew-normal distribution, but the connected area is becoming increasingly broad, and its branches include now many extensions, such as the skew-elliptical families, and some forms of semi-parametric formulations, extending the relevance of the field much beyond the original theme of‘skewness’. The final part of the paper illustrates connections with various areas of application, including selective sampling, models for compositional data, robust methods, some problems in econometrics, non-linear time series, especially in connection with financial data, and more. [ABSTRACT FROM AUTHOR]

We consider the problem of mining from noisy unsupervised data sets. The data point we call noise is an outlier in the current context of data mining, and it has been generally defined as the one locates in low probability regions of an input space. The purpose of the approach for this problem is to detect outliers and to perform efficient mining from noisy unsupervised data. We propose a new iterative sampling approach for this problem, using both model-based clustering and the likelihood given to each example by a trained probabilistic model for finding data points of such low probability regions in an input space. Our method uses an arbitrary probabilistic model as a component model and repeats two steps of sampling non-outliers with high likelihoods (computed by previously obtained models) and training the model with the selected examples alternately. In our experiments, we focused on two-mode and co-occurrence data and empirically evaluated the effectiveness of our proposed method, comparing with two other methods, by using both synthetic and real data sets. From the experiments using the synthetic data sets, we found that the significance level of the performance advantage of our method over the two other methods had more pronounced for higher noise ratios, for both medium- and large-sized data sets. From the experiments using a real noisy data set of protein–protein interactions, a typical co-occurrence data set, we further confirmed the performance of our method for detecting outliers from a given data set. Extended abstracts of parts of the work presented in this paper have appeared in Refs. 1 and 2. [ABSTRACT FROM AUTHOR]

This article presents the results of environmental sciences research into the biological diversity of trees and shrubs in unmanaged forests of Talesh, Iran. 33 woody species, namely 24 trees and nine shrubs, were found in this forests, including Acer cappadocicum, Albizia julibrissin, Carpinus betulus, Cornus australis, Diospyros lotus, Fraxinus coriariifolia, Gleditsia caspica, Juglans regia, Morus alba, Pterocarya fraxinifolia, Vaccinium arctostaphylos, Prunus divaricata, Laurocerasus officinalis, Crataegus microphylla, and Zelkova carpinifolia.

Most existing inductive learning algorithms work under the assumption that their training examples are already tagged. There are domains, however, where the tagging procedure requires significant computation resources or manual labor. In such cases, it may be beneficial for the learner to be active, intelligently selecting the examples for labeling with the goal of reducing the labeling cost. In this paper we present LSS—a lookahead algorithm for selective sampling of examples for nearest neighbor classifiers. The algorithm is looking for the example with the highest utility, taking its effect on the resulting classifier into account. Computing the expected utility of an example requires estimating the probability of its possible labels. We propose to use the random field model for this estimation. The LSS algorithm was evaluated empirically on seven real and artificial data sets, and its performance was compared to other selective sampling algorithms. The experiments show that the proposed algorithm outperforms other methods in terms of average error rate and stability. [ABSTRACT FROM AUTHOR]

In many cost-sensitive environments class probability estimates are used by decision makers to evaluate the expected utility from a set of alternatives. Supervised learning can be used to build class probability estimates; however, it often is very costly to obtain training data with class labels. Active learning acquires data incrementally, at each phase identifying especially useful additional data for labeling, and can be used to economize on examples needed for learning. We outline the critical features of an active learner and present a sampling-based active learning method for estimating class probabilities and class-based rankings. BOOTSTRAP-LV identifies particularly informative new data for learning based on the variance in probability estimates, and uses weighted sampling to account for a potential example's informative value for the rest of the input space. We show empirically that the method reduces the number of data items that must be obtained and labeled, across a wide variety of domains. We investigate the contribution of the components of the algorithm and show that each provides valuable information to help identify informative examples. We also compare BOOTSTRAP-LV with UNCERTAINTY SAMPLING, an existing active learning method designed to maximize classification accuracy. The results show that BOOTSTRAP-LV uses fewer examples to exhibit a certain estimation accuracy and provide insights to the behavior of the algorithms. Finally, we experiment with another new active sampling algorithm drawing from both UNCERTAINTY SAMPLING and BOOTSTRAP-LV and show that it is significantly more competitive with BOOTSTRAP-LV compared to UNCERTAINTY SAMPLING. The analysis suggests more general implications for improving existing active sampling algorithms for classification. [ABSTRACT FROM AUTHOR]

The beamforming process requires a high delay resolution to avoid the deteriorating effects of the delay quantization lobes on the image dynamic range and signal to noise ratio. Wideband transducers require delay resolutions in the order of 1/16 the signal period. If oversampling is used to achieve this timing resolution, a huge data volume has to be acquired and processed in real time. This is usually avoided by sampling just above the Nyquist rate and interpolating to achieve the required delay resolution. However this increases the hardware complexity. Baseband sampling has been alternatively proposed with sampling rates as low as the transducer frequency or even lower. This approach uses two A/D converters and processing chains for every channel, thus doubling the hardware requirements. Quadrature sampling can be used instead with a single A/D converter, but the sampling rate must be a multiple of four times the transducer frequency, decreasing the application flexibility. Furthermore, it produces relatively high errors in the detected envelope if wideband transducers are used.This work presents a new approach, the selective sampling technique (SST), which keeps the lowest sampling rate required by the imaging process or the signal bandwidth (whatever is larger) and, at the same time, provides a high delay resolution to keep the highest image dynamic range. The SST is based on a second order sampling process which, differently from the mentioned approaches, does not pose any constraints in the time interval between samples and produce lower errors in the detected envelope. The hardware requirements are low (a single A/D converter and processing chain for every transducer element), working at the lowest data rate compatible with the Nyquist criterion, thus reducing the data bandwidth. Furthermore, the sampling points can be also freely chosen, so that the SST simplify the usually required scan conversion process to a simple linear interpolation easily carried out by software in real-time. [Copyright &y& Elsevier]

A long-standing goal in the realm of Machine Learning is to minimize sample-complexity, i.e. to reduce as much as possible the number of examples used in the course of learning. The Active Learning paradigm is one such method aimed at achieving this goal by transforming the learner from a passive participant in the information gathering process to an active one. Vaguely speaking, the learner tries to minimize the number of labeled instances used in the course of learning, relaying also on unlabelled instances in order to acquire the needed information whenever possible. The reasoning comes from many real-life problems where the teacher''s activity is an expensive resource (e.g. text categorization, part of speech tagging). The Query By Committee (QBC) (Seung et al., Query by committee, Proceedings of the Fifth Workshop on Computational Learning theory, Morgan Kaufman, San Mateo, CA, 1992, pp. 287–294) is an Active Learning algorithm acting in the Bayesian model of concept learning, (Haussler et al., Mach. Learning 14 (1994) 83) i.e. it assumes that the concept to be learned is chosen according to some fixed and known distribution. Trying to apply the QBC algorithm for learning the class of linear separators, one faces the problem of implementing the mechanism of sampling hypotheses (the Gibbs oracle). The major problem is computational-complexity, since the straightforward Monte Carlo method takes exponential time. In this paper we address the problems involved in the implementation of such a mechanism. We show how to convert them to questions about sampling from convex bodies or approximating the volume of such bodies. Similar problems have recently been solved in the field of computational geometry based on random walks. These techniques enable us to device efficient implementations of the QBC algorithm. We also give few improvements and corrections to the QBC algorithm, the most important one is dropping the Bayes assumption when the concept classes possess a sort of symmetry property (which holds for linear separators). We draw attention to a useful geometric lemma which bounds the maximal radius of a ball contained in a convex body. Finally, this paper exhibits a connection between random walks and certain Machine Learning notions such as ϵ-net and support vector machines. [Copyright &y& Elsevier]

Support vector machines have met with significant success in numerous real-world learning tasks. However, like most machine learning algorithms, they are generally applied using a randomly selected training set classified in advance. In many settings, we also have the option of using pool-based active learning. Instead of using a randomly selected training set, the learner has access to a pool of unlabeled instances and can request the labels for some number of them. We introduce a new algorithm for performing active learning with support vector machines, i.e., an algorithm for choosing which instances to request next. We provide a theoretical motivation for the algorithm using the notion of a version space. We present experimental results showing that employing our active learning method can significantly reduce the need for labeled training instances in both the standard inductive and transductive settings. [ABSTRACT FROM AUTHOR]

Although metalworking fluids have been widely used throughout industry for decades, occupational exposures to metalworking fluid aerosols and their constituents have not been well characterized. This article describes an exposure assessment for a study of metalworking fluid aerosols and acute respiratory effects. This exposure assessment was unique in its inclusion of multiple exposure measures relevant to a complex environment, and extensive personal sampling for bacteria and endotoxin. The specific objectives were to: (1) obtain indices of personal exposure to metalworking fluid aerosols in an automotive transmission plant, either directly (by sampling) or indirectly (by estimation), and (2) identify and adjust for sources of error in exposure/dose measures, where possible. No prior studies have characterized personal exposures to metalworking fluid aerosols so extensively. Exposure data were obtained during a pilot phase and three principal rounds of data collection over a 15-month period in conjunction with spirometric testing. Subjects worked in one of two machining departments, Case and Valve Body, or in a comparison department, Final Assembly. The primary exposure measures for this study were thoracic fraction particulate, thoracic fraction bacteria (viable plus non-viable), and total endotoxin. Mean personal air concentrations of thoracic particulate across all study rounds were 0.13 mg/m[sup 3] in Final Assembly, 0.32 mg/m[sup 3] in Valve Body, and 0.56 mg/m[sup 3] in Case. Average personal exposures to thoracic fraction bacteria were 0.38 bacteria/cc in Final Assembly, 0.87 bacteria/cc in Valve Body, and 2.66 bacteria/cc in Case. Average personal endotoxin measurements, collected in Round 3 of the study, were 16.4 endotoxin units (EU /m[sup 3] in Assembly, 34.7 EU/m[sup 3] in Valve Body, and 234 EU/m[sup 3] in Case. Sump fluid contained on the order of 10[sup 8] bacteria/ml, and 10[sup 4]–10[sup 5] EU/ml. Air concentrations of thoracic particulate, thoracic bacteria, and total endotoxin were highly correlated in met alworking operations. Thus, reducing airborne particulate levels should also reduce ambient bacteria and endotoxin, which are suspect agents of respiratory impairment. The elevated endotoxin levels in Final Assembly were unexpected, and suggest an independent source of endotoxin contamination in this department. [ABSTRACT FROM AUTHOR]

We analyze the “query by committee” algorithm, a method for filtering informative queries from a random stream of inputs. We show that if the two-member committee algorithm achieves information gain with positive lower bound, then the prediction error decreases exponentially with the number of queries. We show that, in particular, this exponential decrease holds for query learning of perceptrons. [ABSTRACT FROM AUTHOR]