Ensemble Performance Research Articles

콘서트홀 무대에서 음향지표와 앙상블 연주와의 상관관계

This paper investigates stage acoustics of a concert hall for ensemble performance in terms of stage support parameters. Although more than 30 years of studies about stage acoustics, only parameter physically measurable for stage acoustics is ST1(Gade,1989, STEarly). STEarly is based on the condition of anechoic chamber and only considered by solo instrumental players' subjective evaluation which means solo singers and ensembles are not considered. So this investigation was set by the condition of ensemble on stage of a concert hall and subjective evaluation was conducted to derive a parameter for the condition of stage acoustics for ensemble. For this investigation, a concert hall with large stage area was selected. Objective measurements were carried out according to stage support evaluation. The sound field on the stage was evaluated using stage and audience acoustical parameters. According to the positions at stage, different acoustical characteristics were found. Higher STEarly and lower IACCE3 were measured near side and rear stage walls. In case of conductor/soloist vs. orchestra player, objective measurement of acoustical parameters shows that early sound fields depend on the source-to-receiver distance but late sound fields depend on the location of side and rear walls. In addition, in-situ subjective evaluation of the trio instrumentalists was carried out. Results show that 'ensemble' is favored by higher early reflection of sounds and sound diffusion affecting 'blending' of their performance.

Short‐range probabilistic forecasting of convective risks for aviation based on a lagged‐average‐forecast ensemble approach

ABSTRACTAn hourly initialized numerical weather prediction model, AROME‐NWC, optimized for nowcasting purposes was used in this study to predict the probabilities of occurrence of convective aviation risks by generating an ensemble of time‐lagged forecasts. The objective is the prediction of echotop and reflectivity maximum based on simulated 3D radar reflectivity columns. Forecasts were postprocessed using an upscaling of the model output fields in order to account for uncertainties in horizontal positions. Simulated radar reflectivities were bias corrected using a quantile‐to‐quantile mapping resulting in an improvement of the ensemble performance. A lagged‐average‐forecast ensemble was then constructed in order to blend mesoscale deterministic and ensemble forecasts, using numerical weather prediction systems that will soon be available in real time. The probabilities of reflectivities predicted by the ensemble are shown to have objective value at thresholds that are meaningful for air traffic control. Possible applications for aviation management purposes are discussed.

Extensive Tonotopic Mapping across Auditory Cortex Is Recapitulated by Spectrally Directed Attention and Systematically Related to Cortical Myeloarchitecture.

Auditory selective attention is vital in natural soundscapes. But it is unclear how attentional focus on the primary dimension of auditory representation—acoustic frequency—might modulate basic auditory functional topography during active listening. In contrast to visual selective attention, which is supported by motor-mediated optimization of input across saccades and pupil dilation, the primate auditory system has fewer means of differentially sampling the world. This makes spectrally-directed endogenous attention a particularly crucial aspect of auditory attention. Using a novel functional paradigm combined with quantitative MRI, we establish in male and female listeners that human frequency-band-selective attention drives activation in both myeloarchitectonically estimated auditory core, and across the majority of tonotopically mapped nonprimary auditory cortex. The attentionally driven best-frequency maps show strong concordance with sensory-driven maps in the same subjects across much of the temporal plane, with poor concordance in areas outside traditional auditory cortex. There is significantly greater activation across most of auditory cortex when best frequency is attended, versus ignored; the same regions do not show this enhancement when attending to the least-preferred frequency band. Finally, the results demonstrate that there is spatial correspondence between the degree of myelination and the strength of the tonotopic signal across a number of regions in auditory cortex. Strong frequency preferences across tonotopically mapped auditory cortex spatially correlate with R1-estimated myeloarchitecture, indicating shared functional and anatomical organization that may underlie intrinsic auditory regionalization.SIGNIFICANCE STATEMENT Perception is an active process, especially sensitive to attentional state. Listeners direct auditory attention to track a violin's melody within an ensemble performance, or to follow a voice in a crowded cafe. Although diverse pathologies reduce quality of life by impacting such spectrally directed auditory attention, its neurobiological bases are unclear. We demonstrate that human primary and nonprimary auditory cortical activation is modulated by spectrally directed attention in a manner that recapitulates its tonotopic sensory organization. Further, the graded activation profiles evoked by single-frequency bands are correlated with attentionally driven activation when these bands are presented in complex soundscapes. Finally, we observe a strong concordance in the degree of cortical myelination and the strength of tonotopic activation across several auditory cortical regions.

Novel Use of Machine Learning for Predicting Radiation Esophagitis in Locally Advanced Stage II-III Non–small Cell Lung Cancer

Radiation-induced esophagitis (RE) is an observed side effect of radiation therapy, can be dose limiting, and is associated with worse overall survival. Currently, few factors beyond esophageal radiation dose are known to be associated with RE. We hypothesize that the use of machine learning (ML) tools can identify factors associated with the development of RE. In this study, we used a dataset from a cohort of 203 consecutive stage II-III locally advanced NSCLC patients, of which 11.3% developed grade≥3 RE. Patients were treated between 2008 and 2016 with IMRT or proton therapy to a median dose of 66.6/1.8 Gy (range 60-80 Gy in 1.8-2.5 Gy fractions). We evaluated 32 continuous and categorical features per patient grouped into risk factors, comorbidities, pretreatment imaging, stage, histology, radiation treatment, chemotherapy, and dosimetry. Univariate analysis was performed using optimally trained decision stumps to determine statistically significant features and their corresponding RE thresholds. Multivariate analysis was also carried out using sequential forward floating selection (SFFS) with decision trees for feature selection. To assess the combined capacity of prediction of the features, a sampling/boosting approach known as RUSBoost was used, which allows the training of strong classifiers from skewed datasets. Balanced cross-validation was used to identify optimal series and thresholds of discriminative variables that map the feature values to the possible presence of RE. The area under the receiver operating characteristic curve (AUC) was used to measure the combined classification performance of the selected features. Univariate analysis showed that esophagus maximum dose >65.1 Gy (p=0.001), lung V20>28% (p=0.02) and heart V5>34% (p=0.04) could consistently predict the presence of RE and are significantly correlated (p<0.05). Moreover, the heart mean, lung mean, and total delivered doses, as well as the T stage, agents/drugs, heart V60, and pre-FEV1 were statistically significant (all p<0.05), while age at diagnosis and esophagus mean dose were marginally significant (both p=0.05) to predict RE. On multivariate analysis, the feature selection using SFFS chose the esophagus maximum dose in 45 and the heart mean dose in 53 of 100 experiment trials. The AUC of the combined classification performance of the RUSBoost ensemble with 250 trees was 0.62 for a 95% confidence interval with confidence range 0.52-0.72. This is the largest report to date of ML applied to predict RE. We found that ML allows for the identification of features and thresholds predictive of RE. These findings can inform treatment planning for radiation oncologists and may ultimately allow for personalized treatment delivery and a reduction in patient morbidity. Prospective data validating these findings are needed.

The impact of horizontal resolution and ensemble size for convective‐scale probabilistic forecasts

The relative benefits of ensemble size and model resolution are investigated within the AROME–France convective‐scale ensemble prediction system (EPS), which operationally runs 12 perturbed members at 2.5 km horizontal resolution. This baseline configuration is compared with two auxiliary ensemble experiments, which are run at resolutions of 2.5 and 1.3 km, with 34 and 12 members respectively. In addition, post‐processing techniques including neighbourhood approaches and time‐lagging are examined as potential alternatives to increase the sample size at a lower computational cost.Probabilistic verification of the raw EPS outputs indicates that increasing model resolution is more beneficial at very short ranges, whereas increasing the ensemble size has a larger impact at longer forecast ranges, as predictability decreases and more members are required to sample the larger uncertainty better. The neighbourhood processing confirms these conclusions and is shown to improve precipitation forecasts of all EPS configurations significantly, especially the smaller‐size ensembles. Hence, it can be considered as a viable substitute to running additional AROME members for the first 24 h of forecasts.Time‐lagging of three successive ensemble productions also appears to be a competitive approach to improve the ensemble forecast skill, particularly at very short ranges, where AROME‐EPS is known to be underdispersive. The performance of the time‐lagged 36 member ensemble is generally close to or even better than the performance of the single 34 member ensemble for the whole forecast range and for different surface weather variables.Overall, given the relative costs and skills of the different EPS configurations, we suggest that resources should primarily be spent on increasing ensemble size, by combining both additional members and post‐processing methods. In addition, the performance of the 1.3 km ensemble should be examined further for explicit high‐impact weather and with an appropriate tuning of physics and surface perturbations.

Assessment of CORDEX-SA experiments in representing precipitation climatology of summer monsoon over India

The present work assesses the performance of 11 regional climate simulations in representing the precipitation patterns of summer monsoon over India for the period 1970–2005. These simulations have been carried out under Coordinated Regional Climate Downscaling Experiment–South Asia (CORDEX-SA) project. The regional climate models (RCMs) have been inter-compared as well as evaluated against the observation to identify the common weaknesses and differences between them. For this, a number of statistical analysis has been carried out to compare the model precipitation field with the corresponding observation. Model uncertainty has been also evaluated through bias studies and analysis of the spread in the ensemble mean (hereafter, ensemble). The models which perform better than the rest are identified and studied to look for any improvement in the ensemble performance. These better performing experiments (best RCM experiments) are further assessed over the monsoon core region (MCR) of India. This has been done to understand how well the models perform in a spatially homogeneous zone of precipitation which is considered to be a representative region of Indian summer monsoon characteristics. Finally, an additional analysis has been done to quantify the skill of models based on two different metrics—performance and convergence including a combination of the two. The experiment with regional model RegCM4 forced with the global model GFDL-ESM2M shows the highest combined mean skill in capturing the seasonal mean precipitation. In general, a significant dry bias is found over a larger part of India in all the experiments which seems most pronounced over the central Indian region. Ensemble on an average tends to outperform many of the individual experiments with bias of smaller magnitude and an improved spatial correlation compared with the observation. Experiments which perform better over India improve the results but only slightly in terms of agreement among experiments and bias.

Pruning the Ensemble of ANN Based on Decision Tree Induction

Ensemble learning is a powerful approach for achieving more accurate predictions compared with single classifier. However, this powerful classification ability is achieved at the expense of heavy storage requirements and computational burdens on the ensemble. Ensemble pruning is a crucial step for the reduction of the predictive overhead without worsening the performance of original ensemble. This paper suggests an efficient and effective ordering-based ensemble pruning based on the induction of decision tree. The suggested method maps the dataset and base classifiers to a new dataset where the ensemble pruning can be transformed to a feature selection problem. Furthermore, a set of accurate, diverse and complementary base classifiers can be selected by the induction of decision tree. Moreover, an evaluation function that deliberately favors the candidate sub-ensembles with an improved performance in classifying low margin instances has also been designed. The comparative experiments on 24 benchmark datasets demonstrate the effectiveness of our proposed method.

MRMR-based ensemble pruning for facial expression recognition

Facial expression recognition (FER) can assist the interaction between humans and devices. The combination of FER and ensemble learning can usually improve final recognition results. However, in many cases, the produced ensemble classifiers often contain many redundant members, and those components bring potential side effects to final results. Previous studies have illustrated that a more compact subset of a classifier pool shows better performance than original classifier pool. Furthermore, the compacted subset reduces storage space and decreases computation complexity. This paper proposes a maximum relevance and minimum redundancy-based ensemble pruning (MRMREP) method that treats prediction results as features, and extends the feature selection method to the ensemble classifier reduction problem to obtain a more representative subset. This novel method ordered all base classifiers according to two important factors: the correlation between target labels and predictions, and the redundancy between classifiers. The final ensemble performance was evaluated by comparing our method with other ensemble pruning methods, and superior results were obtained on the FER2013, JAFFE, and CK + databases.

Reaction: Technological Aspects of Molecular Electronics

Dr. Emanuel Lortscher received his PhD with distinction (summa cum laude) in 2006 from the University of Basel in Switzerland and a master in physics in 2003 from the Swiss Federal Institute of Technology. He has been a research staff member at IBM Research since 2008. He has received various prizes, including the Swiss Physical Society Award for Applied Physics in 2007 and the Faculty Prize of the University of Basel in the same year. His current research focuses on fabrication, electronics, optics, and chemistry at the nanoscale for applications in novel computing applications, biology, medicine, and the Internet of Things.

Analysis on Ensemble of Regenerating Codes and Replication

This letter studies the performance of ensemble of Regenerating codes and Replication (RR). It theoretically analyzes the storage-bandwidth tradeoff curve of RR, and attains two extremal cases, corresponding to minimum storage cost RR (MSRR) and minimum repair bandwidth RR (MBRR), respectively. Compared with the case of independently applying regenerating codes, the evaluation results indicate, with extra storage cost, MSRR has better average repair bandwidth and average disk I/O, and MBRR has less average disk I/O cost and almost the same level of average repair bandwidth, i.e., when satisfying the data available probability $P=0.9999$ , MSRR achieves at least 5.6% and 50%, and MBRR achieves at least 23.7% and 1.3% less disk I/O and repair bandwidth consumption in evaluation data sets, respectively.

Online Orchestra: Connecting remote communities through music

Abstract Online Orchestra is a telematic performance project, aimed at enabling young and amateur musicians in geographically remote locations to make music together over the Internet. This article describes the contexts out of which the project emerged, including an overview of the benefits of ensemble performance, and a survey of precedent telematic performance projects. It goes on to describe how the starting premises of Online Orchestra respond to these contexts and ends with a summary of Online Orchestra’s approach, and its key findings. The article describes in particular how many recent telematic performance projects rely on specialist networks and equipment, and that alternative design solutions are necessary, and possible, in order to reach young and amateur musicians in their own remote locations.

Jaunimo etnokultūrinis ugdymas aukštųjų mokyklų muzikiniuose ansambliuose

Straipsnyje analizuojamas Lietuvos ir šalių kaimynių (Latvijos ir Lenkijos), susietų su Lietuva kultūriniais ir istoriniais ryšiais, aukštųjų mokyklų jaunimo požiūris į muzikavimą etnokultūriniame ansamblyje kaip priemonę, padedančią ugdyti etninį ir kultūrinį tapatumą (tautinį identitetą). Atliktas empirinis tyrimas* atskleidžia muzikinės veiklos etnokultūriniame ansamblyje, kaip vienos iš neformaliojo švietimo formų, pasirinkimo motyvus, etnokultūrinio ugdymo svarbą ir asmeninę reikšmę jaunimo gyvenime.

The best of two worlds: Balancing model strength and comprehensibility in business failure prediction using spline-rule ensembles

Numerous organizations and companies rely upon business failure prediction to assess and minimize the risk of initiating business relationships with partners, clients, debtors or suppliers. Advances in research on business failure prediction have been largely dominated by algorithmic development and comparisons led by a focus on improvements in model accuracy. In this context, ensemble learning has recently emerged as a class of particularly well-performing methods, albeit often at the expense of increased model complexity. However, in practice, model choice is rarely based on predictive performance alone. Models should be comprehensible and justifiable to assess their compliance with common sense and business logic, and guarantee their acceptance throughout the organization. A promising ensemble classification algorithm that has been shown to reconcile performance and comprehensibility are rule ensembles. In this study, an extension entitled spline-rule ensembles is introduced and validated in the domain of business failure prediction. Spline-rule ensemble complement rules and linear terms found in conventional rule ensembles with smooth functions with the aim of better accommodating nonlinear simple effects of individual features on business failure. Experiments on a large selection of 21 datasets of European companies in various sectors and countries (i) demonstrate superior predictive performance of spline-rule ensembles over a set of well-established yet powerful benchmark methods, (ii) show the superiority of spline-rule ensembles over conventional rule ensembles and thus demonstrate the value of the incorporation of smoothing splines, (iii) investigate the impact of alternative term regularization procedures and (iv) illustrate the comprehensibility of the resulting models through a case study. In particular, the ability of the technique to reveal the extent and the way in which predictors impact business failure, and if and how variables interact, are exemplified.

Moisture transmission behaviour of individual component and multi-layered fabric with sweat and pure water

Moisture transport through multilayer clothing system is one of the most important aspect governing the comfort and performance of multi-layered clothing system. In the present study, properties critical to functioning of outdoor apparels were tested using sweat and pure water for both multi-layered and single layer components i.e., polyester knitted fabric as inner layer (next to skin), polyester fleece/polyester spacer as middle layer and PU coated nylon as outer layer. Water vapour transmission rate of fabrics when tested with sweat solution was significantly lower as compared to water. Fabric with sweat also showed lower drying rate than with water. For sweat solution, absorption rate is higher for both inner and outer surface as compared to pure water for single layer fabrics. However, the wetting time (inversely related to spreading speed) is higher at inner surface and lower at outer surface. In general, the performance of multi-layered ensemble is dominated by the impact of PU coated layer. Multi-layered structure with spacer fabric as middle layer exhibits better water vapour transmission rate, overall moisture management coefficient and drying rate as compared to multi-layered structure with fleece fabric.

Margin &amp; diversity based ordering ensemble pruning

Abstract Ensemble pruning is a technique used to improve ensemble performance and reduce the ensemble size by selecting an optimal or sub-optimal subset as the final ensemble for prediction. In this research, using example margin and ensemble diversity, we prove that the ensemble pruning method should focus more on the following two factors: (1) examples with small absolute margin and (2) classifiers that correctly classify more examples and contribute larger diversity. Based on this principle, we propose a novel metric called the margin & diversity based measure (MDM) to explicitly evaluate the importance of individual classifiers. By incorporating ensemble members in a decreasing order based on the MDM, sub-ensembles are formed such that users can select the top T ensemble members for predictions. Compared to the original ensemble and other state-of-the-art ensemble pruning methods, the proposed method shows better performance in terms of accuracy.

An evaluation of the performance of a WRF multi-physics ensemble for heatwave events over the city of Melbourne in southeast Australia

Appropriate choice of physics options among many physics parameterizations is important when using the Weather Research and Forecasting (WRF) model. The responses of different physics parameterizations of the WRF model may vary due to geographical locations, the application of interest, and the temporal and spatial scales being investigated. Several studies have evaluated the performance of the WRF model in simulating the mean climate and extreme rainfall events for various regions in Australia. However, no study has explicitly evaluated the sensitivity of the WRF model in simulating heatwaves. Therefore, this study evaluates the performance of a WRF multi-physics ensemble that comprises 27 model configurations for a series of heatwave events in Melbourne, Australia. Unlike most previous studies, we not only evaluate temperature, but also wind speed and relative humidity, which are key factors influencing heatwave dynamics. No specific ensemble member for all events explicitly showed the best performance, for all the variables, considering all evaluation metrics. This study also found that the choice of planetary boundary layer (PBL) scheme had largest influence, the radiation scheme had moderate influence, and the microphysics scheme had the least influence on temperature simulations. The PBL and microphysics schemes were found to be more sensitive than the radiation scheme for wind speed and relative humidity. Additionally, the study tested the role of Urban Canopy Model (UCM) and three Land Surface Models (LSMs). Although the UCM did not play significant role, the Noah-LSM showed better performance than the CLM4 and NOAH-MP LSMs in simulating the heatwave events. The study finally identifies an optimal configuration of WRF that will be a useful modelling tool for further investigations of heatwaves in Melbourne. Although our results are invariably region-specific, our results will be useful to WRF users investigating heatwave dynamics elsewhere.

SOM-based partial labeling of imbalanced data stream

Data streams are found in many large-scale systems such as security, finance, and internet. In many of the data streams, the class distribution is imbalanced, and hence most of the traditional classification modeler fails to produce high accuracy for the samples from minority class. In addition, data streams are changing and the model should be updated to maintain the classification performance. However, obtaining the true class labels of the data samples is not an easy task since labeling process is extremely time-consuming and very often class labels are not available immediately after classification. The goal of this research is to reduce the labeling for an imbalanced data stream, and to produce high classification performance when compared to fully labeling setting. In an imbalanced data stream, the challenging part is to find and label minority class samples. In this paper, we propose RLS-SOM (Reduced labeled Samples-Self Organizing Map) framework for classification of the imbalanced data stream in a non-stationary environment. RLS-SOM locates the minority class samples in the feature space using SOM. It maintains an ensemble of the classifiers and builds a new model when the changes occur, using only partial labeled samples. In RLS-SOM, the classification results are obtained from the ensemble, as well as each individual model in the ensemble. An individual model classification results are selected over ensemble results, if its performance is higher than the ensemble's performance. This comparison is performed to improve the performance as there may be one model in the ensemble that produces higher performance than the ensemble. Our experimental results demonstrate that RLS-SOM obtains higher performance when it is compared with several partially labeling techniques over benchmark data sets. In addition, the experimental results with other state of the art fully labeling methods such as UCB, SERA, SEA, and Learn++.CDS shows RLS-SOM maintains equivalent classification performance by using 10–30% labeling, on average.

Global evaluation of runoff from 10 state-of-the-art hydrological models

Abstract. Observed streamflow data from 966 medium sized catchments (1000–5000 km2) around the globe were used to comprehensively evaluate the daily runoff estimates (1979–2012) of six global hydrological models (GHMs) and four land surface models (LSMs) produced as part of tier-1 of the eartH2Observe project. The models were all driven by the WATCH Forcing Data ERA-Interim (WFDEI) meteorological dataset, but used different datasets for non-meteorologic inputs and were run at various spatial and temporal resolutions, although all data were re-sampled to a common 0. 5° spatial and daily temporal resolution. For the evaluation, we used a broad range of performance metrics related to important aspects of the hydrograph. We found pronounced inter-model performance differences, underscoring the importance of hydrological model uncertainty in addition to climate input uncertainty, for example in studies assessing the hydrological impacts of climate change. The uncalibrated GHMs were found to perform, on average, better than the uncalibrated LSMs in snow-dominated regions, while the ensemble mean was found to perform only slightly worse than the best (calibrated) model. The inclusion of less-accurate models did not appreciably degrade the ensemble performance. Overall, we argue that more effort should be devoted on calibrating and regionalizing the parameters of macro-scale models. We further found that, despite adjustments using gauge observations, the WFDEI precipitation data still contain substantial biases that propagate into the simulated runoff. The early bias in the spring snowmelt peak exhibited by most models is probably primarily due to the widespread precipitation underestimation at high northern latitudes.

Tree-based classifier ensembles for early detection method of diabetes: an exploratory study

Diabetes is a lifestyle-driven disease which has become a critical health issue worldwide. In this paper, we conduct an exploratory study about early detection method of diabetes mellitus using various ensemble learning techniques. Eight tree-based machine learning algorithms, i.e. classification and regression tree, decision tree (C4.5), reduced error pruning tree, random tree, naive Bayes tree, functional tree, best-first decision tree and logistic model tree are employed as a base classifier in five different ensembles, i.e. bagging, boosting, random subspace, DECORATE, and rotation forest. The performance of ensembles and base classifiers are thoroughly benchmarked on three real-world datasets in term of area under receiver operating characteristic curve metric. Finally, we assess the performance differences among the classifiers using several statistical significant tests. We contribute to the existing literature regarding an extensive benchmark of tree-based classifier ensembles for early detection method of diabetes disease.

Exploring diversity in ensemble classification: Applications in large area land cover mapping

Ensemble classifiers, such as random forests, are now commonly applied in the field of remote sensing, and have been shown to perform better than single classifier systems, resulting in reduced generalisation error. Diversity across the members of ensemble classifiers is known to have a strong influence on classification performance - whereby classifier errors are uncorrelated and more uniformly distributed across ensemble members. The relationship between ensemble diversity and classification performance has not yet been fully explored in the fields of information science and machine learning and has never been examined in the field of remote sensing. This study is a novel exploration of ensemble diversity and its link to classification performance, applied to a multi-class canopy cover classification problem using random forests and multisource remote sensing and ancillary GIS data, across seven million hectares of diverse dry-sclerophyll dominated public forests in Victoria Australia. A particular emphasis is placed on analysing the relationship between ensemble diversity and ensemble margin - two key concepts in ensemble learning. The main novelty of our work is on boosting diversity by emphasizing the contribution of lower margin instances used in the learning process. Exploring the influence of tree pruning on diversity is also a new empirical analysis that contributes to a better understanding of ensemble performance. Results reveal insights into the trade-off between ensemble classification accuracy and diversity, and through the ensemble margin, demonstrate how inducing diversity by targeting lower margin training samples is a means of achieving better classifier performance for more difficult or rarer classes and reducing information redundancy in classification problems. Our findings inform strategies for collecting training data and designing and parameterising ensemble classifiers, such as random forests. This is particularly important in large area remote sensing applications, for which training data is costly and resource intensive to collect.