Proportion Of Instances Research Articles

Gaze Scanning at Street Crossings by Pedestrians With Homonymous Hemianopia With and Without HemispatialNeglect.

To investigate compensatory gaze-scanning behaviors during street crossings by pedestrians with homonymous hemianopia (HH) and hemispatial neglect (HSN). Pedestrians with right homonymous hemianopia (RHH) and left homonymous hemianopia without (LHH) and with left spatial-neglect (LHSN) walked on city streets wearing a gaze-tracking system that also captured scene videos. Street-crossing instances were manually annotated, and horizontal gaze scan of magnitude ≥20° and scanning rates were compared within-subject, between the side of the hemifield loss (BlindSide) and the other side (SeeingSide). Proportion of instances with scans to both the left and the right side at nonsignalized crossings (indicative of safe scanning behavior) were compared among the three subject groups. Data from 19 participants (6 LHH, 7 RHH, and 6 with mild [4] or moderate [2] LHSN), consisting of 521 street-crossing instances of a total duration of 201 minutes and 5375 gaze scans, were analyzed. The overall gaze magnitude (mean [95% confidence interval (CI)]) was significantly larger toward the BlindSide (40.4° [39.1°-41.9°]) than the SeeingSide (36° [34.8°-37.3°]; P < 0.001). The scanning rate (mean [95% CI] scans/min) toward the BlindSide (14 [12.5-15.6]) was significantly higher than the SeeingSide (11.5 [10.3°-12.9°]; P < 0.001). The scanning rate in the LHSN group (10.7 [8.9-12.8]) was significantly lower than the LHH group (14 [11.6-17.0]; P = 0.045). The proportion of nonsignalized crossings with scans to both sides was significantly lower in LHSN (58%; P = 0.039) and RHH (51%; P = 0.003) than LHH (75%) participants. All groups demonstrated compensatory scanning, making more gaze scans with larger magnitudes to the blind side. Mild to moderate LHSN adversely impacted the scanning rate.

Instance selection for big data based on locally sensitive hashing and double-voting mechanism

The increasing data volumes impose unprecedented challenges to traditional data mining in data preprocessing, learning, and analyzing, it has attracted much attention in designing efficient compressing, indexing and searching methods recently. Inspired by locally sensitive hashing (LSH), divide-and-conquer strategy, and double-voting mechanism, we proposed an iterative instance selection algorithm, which needs to run p rounds iteratively to reduce or eliminate the unwanted bias of the optimal solution by double-voting. In each iteration, the proposed algorithm partitions the big dataset into several subsets and distributes them to different computing nodes. In each node, the instances in local data subset are transformed into Hamming space by l hash function in parallel, and each instance is assigned to one of the l hash tables by the corresponding hash code, the instances with the same hash code are put into the same bucket. And then, a proportion of instances are randomly selected from each hash bucket in each hash table, and a subset is obtained. Thus, totally l subsets are obtained, which are used for voting to select the locally optimal instance subset. The process is repeated p times to obtain p subsets. Finally, the globally optimal instance subset is obtained by voting with the p subsets. The proposed algorithm is implemented with two open source big data platforms, Hadoop and Spark, and experimentally compared with three state-of-the-art methods on testing accuracy, compression ratio, and running time. The experimental results demonstrate that the proposed algorithm provides excellent performance and outperforms three baseline methods.

Representativeness-Based Instance Selection for Intrusion Detection

With the continuous development of network technology, an intrusion detection system needs to face detection efficiency and storage requirement when dealing with large data. A reasonable way of alleviating this problem is instance selection, which can reduce the storage space and improve intrusion detection efficiency by selecting representative instances. An instance is representative not only in its class but also in different classes. This representativeness reflects the importance of an instance. Since the existing instance selection algorithm does not take into account the above situations, some selected instances are redundant and some important instances are removed, increasing storage space and reducing efficiency. Therefore, a new representativeness of instance is proposed and considers not only the influence of all instances of the same class on the selected instance but also the influence of instances of different classes on the selected instance. Moreover, it considers the influence of instances of different classes as an advantageous factor. Based on this representativeness, two instance selection algorithms are proposed to handle balanced and imbalanced data problems for intrusion detection. One is a representative-based instance selection for balanced data, which is named RBIS and selects the same proportion of instances from each class. The other is a representative-based instance selection for imbalanced data, which is named RBIS-IM and selects important majority instances according to the number of instances of the minority class. Compared with other algorithms on the benchmark data sets of intrusion detection, experimental results verify the effectiveness of the proposed RBIS and RBIS-IM algorithms and demonstrate that the proposed algorithms can achieve a better balance between accuracy and reduction rate or between balanced accuracy and reduction rate.

Random Balance ensembles for multiclass imbalance learning

Random Balance strategy (RandBal) has been recently proposed for constructing classifier ensembles for imbalanced, two-class data sets. In RandBal, each base classifier is trained with a sample of the data with a random class prevalence, independent of the a priori distribution. Hence, for each sample, one of the classes will be undersampled while the other will be oversampled. RandBal can be applied on its own or can be combined with any other ensemble method. One particularly successful variant is RandBalBoost which integrates Random Balance and boosting. Encouraged by the success of RandBal, this work proposes two approaches which extend RandBal to multiclass imbalance problems. Multiclass imbalance implies that at least two classes have substantially different proportion of instances. In the first approach proposed here, termed Multiple Random Balance (MultiRandBal), we deal with all classes simultaneously. The training data for each base classifier are sampled with random class proportions. The second approach we propose decomposes the multiclass problem into two-class problems using one-vs-one or one-vs-all, and builds an ensemble of RandBal ensembles. We call the two versions of the second approach OVO-RandBal and OVA-RandBal, respectively. These two approaches were chosen because they are the most straightforward extensions of RandBal for multiple classes. Our main objective is to evaluate both approaches for multiclass imbalanced problems. To this end, an experiment was carried out with 52 multiclass data sets. The results suggest that both MultiRandBal, and OVO/OVA-RandBal are viable extensions of the original two-class RandBal. Collectively, they consistently outperform acclaimed state-of-the art methods for multiclass imbalanced problems.

Robust multiple-instance learning ensembles using random subspace instance selection

Many real-world pattern recognition problems can be modeled using multiple-instance learning (MIL), where instances are grouped into bags, and each bag is assigned a label. State-of-the-art MIL methods provide a high level of performance when strong assumptions are made regarding the underlying data distributions, and the proportion of positive to negative instances in positive bags. In this paper, a new method called Random Subspace Instance Selection (RSIS) is proposed for the robust design of MIL ensembles without any prior assumptions on the data structure and the proportion of instances in bags. First, instance selection probabilities are computed based on training data clustered in random subspaces. A pool of classifiers is then generated using the training subsets created with these selection probabilities. By using RSIS, MIL ensembles are more robust to many data distributions and noise, and are not adversely affected by the proportion of positive instances in positive bags because training instances are repeatedly selected in a probabilistic manner. Moreover, RSIS also allows the identification of positive instances on an individual basis, as required in many practical applications. Results obtained with several real-world and synthetic databases show the robustness of MIL ensembles designed with the proposed RSIS method over a range of witness rates, noisy features and data distributions compared to reference methods in the literature.

Evaluating authorship distance methods using the positive Silhouette coefficient

AbstractUnsupervised Authorship Analysis (UAA) aims to cluster documents by authorship without knowing the authorship of any documents. An important factor in UAA is the method for calculating the distance between documents. This choice of the authorship distance method is considered more critical to the end result than the choice of cluster analysis algorithm. One method for measuring the correlation between a distance metric and a labelling (such as class values or clusters) is the Silhouette Coefficient (SC). The SC can be leveraged by measuring the correlation between the authorship distance method and the true authorship, evaluating the quality of the distance method. However, we show that the SC can be severely affected by outliers. To address this issue, we introduce the Positive Silhouette Coefficient, given as the proportion of instances with a positive SC value. This metric is not easily altered by outliers and produces a more robust metric. A large number of authorship distance methods are then compared using the PSC, and the findings are presented. This research provides an insight into the efficacy of methods for UAA and presents a framework for testing authorship distance methods.

Best Guess method: A further external validation study and comparison with other methods

Validation of the 'Best Guess' weight estimation method on a geographically divergent external sample of children, plus comparison with APLS and Broselow weight estimation methods. Prospective cross-sectional analytical study at Princess Margaret Hospital Emergency Department. A convenience sample of children aged 0-14 years recruited from May to June 2008. Age, sex, ethnicity, height and actual weight obtained. Agreement between the methods is reported as a comparative mean and distribution of the percentage error, plus the proportion of instances where the error exceeded 20% of the measured weight. A total of 1235 children were included. The 'Best Guess' method was the most accurate, particularly in children aged 1-4 years (mean percentage error +1.69%). In other age groups it overestimated weight, with mean percentage errors ranging from 3.41% to 6.25%. Across all age groups the Broselow method was most precise, with tendency to underestimate weight across age groups with mean percentage errors ranging from -5.28% to -7.24%. The APLS method was least accurate and precise, with mean percentage errors ranging from -12.61% to -17.36%. Net weight underestimation errors exceeding 20% were associated with increased mean body mass index. The Best Guess weight estimation method is accurate, especially in children aged 1-4 years. It moderately overestimates weight in other ages. The Broselow method was more precise, whereas the APLS method was the least accurate and precise of all. The ease of use of the Broselow method argues for greater use in the ED and prehospital setting.

Class noise detection using frequent itemsets

The presence of a substantial number of noisy instances in a given dataset may adversely affect the hypothesis learnt from that data. Removing noisy instances prior to the construction of a classifier has been shown to improve the classification ability of a learner on new data. This paper introduces a novel technique for identifying observations with class noise in a dataset using frequent itemsets. For the given dataset, each instance is assigned a NoiseFactor, indicating a relative likelihood that it contains class noise. A frequent itemset is a set of instances with common attribute values which contains at least as many instances as a user-defined minimum support threshold. Consequently, the set of frequent itemsets contains information related to the structure and dependence between the attributes. Each frequent itemset is assigned a class, based on the proportion of instances within the itemset from each class. Instances that are contained in itemsets that have a large proportion of instances from the other class are identified as noisy. The technique proposed in this paper is analyzed in numerous case studies using real-world software measurement datasets with either inherent or injected noise. A comparison is provided with two well-known techniques for the identification of class noise: Classification Filter and Ensemble Filter. The results demonstrate that this new algorithm is very effective at identifying instances with class noise.

Causal judgment from contingency information: a systematic test of the pCI rule.

Contingency information is information about the occurrence or nonoccurrence of an effect when a possible cause is present or absent. Under the evidential evaluation model, instances of contingency information are transformed into evidence and causal judgment is based on the proportion of relevant instances evaluated as confirmatory for the candidate cause. In this article, two experiments are reported that were designed to test systematic manipulations of the proportion of confirming instances in relation to other variables: the proportion of instances on which the candidate cause is present, the proportion of instances in which the effect occurs when the cause is present, and the objective contingency. Results showed that both unweighted and weighted versions of the proportion-of-confirmatory-instances rule successfully predicted the main features of the results, with the weighted version proving more successful. Other models, including the power PC theory, failed to predict the results.

The use of relative abundance patterns to discriminate among niche apportionment processes

Several different stochastic models of niche apportionment have been proposed to explain the structure of species' relative abundance patterns within communities. Here I deal with the converse issue, i.e., whether it is possible to infer past niche apportionment processes from current relative abundance patterns. Using Monte Carlo methods, I constructed relative abundance patterns for many .communities. of four to 28 species based on five common niche apportionment models (Dominance Decay, Dominance Preemption, MacArthur Fraction, Random Assortment and Random Fraction). The relative abundance patterns of these individual communities were compared by a best-fit test against those expected to emerge over many replications of the different models. The proportion of instances that the past structuring processes were correctly identified was generally quite low for four-species communities, and was inconsistent across niche apportionment models. For 28-species communities, the probability of correct model iden...

Causal judgment from contingency information: The interpretation of factors common to all instances.

Participants made judgments about stimulus materials in which there were 2 possible causes of an outcome. One of these was a common factor, a factor present in all instances presented for judgment, and the other was a positive covariate of the outcome. Instead of interpreting the positive covariate as the cause, participants consistently preferred an interpretation in which the common factor was the cause and the positive covariate enabled the cause to produce its effect. Participants' judgments of both interpretations were predicted by the proportion of instances that were confirmatory for the interpretation and not by ΔP. The findings support a hypothesis that people have a multirole conceptualization of causality including, in addition to the roles of cause and effect, that of an enabler, a factor the presence of which ensures that a thing is in a state of readiness to produce a particular effect.

Causal judgment from contingency information: the interpretation of factors common to all instances.

Participants made judgments about stimulus materials in which there were 2 possible causes of an outcome. One of these was a common factor, a factor present in all instances presented for judgment, and the other was a positive covariate of the outcome. Instead of interpreting the positive covariate as the cause, participants consistently preferred an interpretation in which the common factor was the cause and the positive covariate enabled the cause to produce its effect. Participants' judgments of both interpretations were predicted by the proportion of instances that were confirmatory for the interpretation and not by deltaP. The findings support a hypothesis that people have a multirole conceptualization of causality including, in addition to the roles of cause and effect, that of an enabler, a factor the presence of which ensures that a thing is in a state of readiness to produce a particular effect.

Accuracy of multichannel median filter

The question whether the output of the multichannel median filter should be one of the input vector samples is addressed. While the output is normally one of the inputs in the single-channel case, engineering it to be so in the multichannel case is found to lead to significant error in a large proportion of instances, a situation that should be avoided without good reason.

Cell-Mediated Responses in Dogs With Spontaneous Neoplasms. I. Detection of Cell-Mediated Cytotoxicity by the Chromium-51 Release Assay&lt;xref ref-type="fn" rid="FN1"&gt;2&lt;/xref&gt;

A series of 43 dogs with spontaneous melanomas, sarcomas, or mammary carcinomas were tested for peripheral blood lymphocytotoxicity against a range of allogeneic tumor cells in vitro on one or more occasions during therapy. All tumor-bearer groups contained a proportion of dogs showing a significant 51Cr release with increasing effector-to-target cell ratios. However, cytotoxicity was not restricted for target cells of the same histologic type as that of the effector cell donor. Some unrelated target cells more sensitive to nonspecific effects showed a greater cytotoxicity in some instances. Control effector cells from healthy dogs were nonspecifically cytotoxic for a range of tumor target cells in a similar proportion of instances. Short-term cultures of autochthonous tumor target cells were resistant to lysis in the 51Cr release assay. The findings did not provide evidence for the existence of specific tumor antigens operative to allogeneic 51Cr release cytotoxicity assays.

Quality evaluation and control methods in computer-assisted screening.

A computer-assisted, chronic respiratory disease (CRD) screening project undertook a critical evaluation of the performance of the screening equipment and techniques used in administering the tests and interpreting the measurements. Based upon screening tests of 13,494 adults, the analysis led to several conclusions: the use of trained volunteers to administer the questionnaire was feasible and reproducible within the expectation of such a technique; the “nonexertional wheezing” question was particularly associated with spirometry data; the spirometry system was found to be accurate and stable under screening conditions; technician improvement was observed; the spirometry technique was modified based upon an experiment performed during the course of the screening program; and the dual positive film readings were in agreement in only a small proportion of instances, necessitating a major effort aimed to increase concordance.

Edathamil calcium-disodium (versenate) in treatment of lead poisoning in children.

THE TREATMENT of both acute and chronic lead poisoning in children has been unsatisfactory; acute encephalopathy leading to death or gross mental defect in a high proportion of instances, while the less dramatic forms of infantile plumbism have frequently been followed by failure of orderly cerebral maturation.1 Various methods of treatment, such as surgical decompression, measures directed at depositing circulating lead in the skeleton, deleading with sodium citrate or dimercaprol (BAL), have had little influence on the outcome. Recent publications have suggested that edathamil calcium-disodium (calcium disodium ethylenediaminetetraacetic acid; Ca EDTA; calcium versenate) is an effective agent in safely and rapidly removing lead from the body.* The drug is a chelating agent of the following structual formula: It is soluble in water and the usual watery intravenous solutions such as liquid dextrose and isotonic sodium chloride solution. It forms un-ionized soluble very stable complexes with metallic ions, the two nitrogen

Observations on the inter-relations of pulmonary tuberculosis, influenza and pneumonia

Summary In their own homes the tuberculous suffered attacks of influenza at least as frequently as the population at large. The adult cases of pulmonary tuberculosis in residence at Barrasford Sanatorium and the Tuberculosis Wards at the City Hospital escaped the epidemics almost completely. The isolation of the first-named institution may have accounted for its immunity; on the other hand it is difficult to explain why the hospital patients got off practically “scot free” as 50 per cent, of the nursing and domestic staffs were affected during the 1918–19 epidemic. Post influenzal pseudo-tuberculosis and tuberculous pseudo-influenza caused considerable confusion in diagnosis. Acute influenzal pneumonia was found to be followed by pulmonary tuberculosis in a greater proportion of instances than was lobar pneumonia; the total number of cases, however, was small and there was no evidence that influenza proper was a serious factor in increasing the morbidity of tuberculosis. While 15·9 per cent, of the new sputum-positive cases registered between June 1, 1918, and June 30, 1920, ascribed their pulmonary tuberculosis to influenza it has to be borne in mind that at least 20 per cent, of the population at large suffered from an attack of influenza during the 1918–19 epidemic. Many of the definite cases suffered an aggravation of their disease as a result of an attack of influenza, and two patients with an artificial pneumothorax died of acute influenzal pneumonia. During and immediately after the 1918–19 epidemic the death returns revealed an increase in the number of deaths from pulmonary tuberculosis; this increase can, in part at least, be ascribed to influenza. Since then the death-rate from pulmonary tuberculosis has progressively declined and that of 1922 was the lowest ever recorded in the city.

MANUAL AND INSTRUMENTAL DILATATION OF THE PREGNANT AND PARTURIENT CERVIX.

Most of the graver obstetric operations find their necessity in the difficulty or impossibility of overcoming two barriers to delivery by the natural passages. The first of these is the resistance offered by the bony pelvic inlet and the second the obstruction of the cervix. Leaving out of consideration for the moment the causes of fetal dystocia, I believe that, broadly speaking, the most important factor in maternal dystocia is in the large proportion of instances of this second barrier to the progress of delivery, namely, the barrier of the cervix. Unlike the barrier presented by a contracted pelvis, this barrier of the cervix is always present; and although it is true that in physiologic labor it is readily and safely removed by the natural efforts, it is, I firmly believe, equally true that unfortunate results for both mother and child are far oftener the consequence of non-dilatation or unsatisfactory

PAROXYSMAL TACHYCARDIA: ITS RELATION TO EXOPHTHALMIC GOITER.

It is my wish to emphasize the relationship apparently existing between certain cases of paroxysmal tachycardia and exophthalmic goiter, and to suggest that a more thorough examination of such cases may serve to trace the connection in a considerable proportion of instances. The more I see of this symptomatic disturbance, the less inclined am I to regard it as a definite entity or deserving of special caption in the category of disease. A superficial survey of the literature shows some 160 references to the subject, and as yet there has been no opportunity to observe what proportion of the cases are regarded by their reporters to have been due to exophthalmic goiter, though it is apparent that the percentage is small. It is hoped that a thorough resumé of the subject may be embodied in a future note. The present communication is actuated by two recent cases in my practice,

On Duplicitas Anterior in an Early Chick Embryo

The literature of Duplicity in Birds affords, out of a total of about ninety-five recorded cases of multiple formations of all kinds on a single blastoderm, from the stage of the primitive streak to the fourth day of incubation, only a small proportion of instances of “duplicitas anterior.” Dareste (i.) in his atlas figures three; Gerlach (ii.) adds representations of three others—one case of his own, a second originally described by Ahlfeld, and a third by Reichert; Klaussner (iii.) gives a seventh case; and Bianchi (iv.) describes a monstrous embryo at a later stage (1°5 cm. in length).Most observers have been content with the partial information derived from the study of the whole object, and only three embryos of this class, which have been studied in serial sections, have been described:—1st. Erich Hoffman's (v.) with three somites.2nd. Mitrophanow's (vi.) with six somites.3rd. Kaestner's (vii.) with seven somites.