Construction of neurofuzzy models for imbalanced data classification

We propose a new class of neurofuzzy construction algorithms with the aim of maximizing generalization capability specifically for imbalanced data classification problems based on leave-one-out (LOO) cross validation. The algorithms are in two stages, first an initial rule base is constructed based on estimating the Gaussian mixture model with analysis of variance decomposition from input data; the second stage carries out the joint weighted least squares parameter estimation and rule selection using orthogonal forward subspace selection (OFSS)procedure. We show how different LOO based rule selection criteria can be incorporated with OFSS, and advocate either maximizing the leave-one-out area under curve of the receiver operating characteristics, or maximizing the leave-one-out Fmeasure if the data sets exhibit imbalanced class distribution. Extensive comparative simulations illustrate the effectiveness of the proposed algorithms.

Unified framework for construction of rule based classification systems

Automatic generation of classification rules has been an increasingly popular technique in commercial applications such as Big Data analytics, rule based expert systems and decision making systems. However, a principal problem that arises with most methods for generation of classification rules is the overfit-ting of training data. When Big Data is dealt with, this may result in the generation of a large number of complex rules. This may not only increase computational cost but also lower the accuracy in predicting further unseen instances. This has led to the necessity of developing pruning methods for the simplification of rules. In addition, classification rules are used further to make predictions after the completion of their generation. As efficiency is concerned, it is expected to find the first rule that fires as soon as possible by searching through a rule set. Thus a suit-able structure is required to represent the rule set effectively. In this chapter, the authors introduce a unified framework for construction of rule based classification systems consisting of three operations on Big Data: rule generation, rule simplification and rule representation. The authors also review some existing methods and techniques used for each of the three operations and highlight their limitations. They introduce some novel methods and techniques developed by them recently. These methods and techniques are also discussed in comparison to existing ones with respect to efficient processing of Big Data.

Advanced feature selection methods in multinominal dementia classification from structural MRI data

Recent studies showed that features extracted from brain MRIs can well discriminate Alzheimer’s disease from Mild Cognitive Impairment. This study provides an algorithm that sequentially applies advanced feature selection methods for findings the best subset of features in terms of binary classification accuracy. The classifiers that provided the highest accuracies, have been then used for solving a multi-class problem by the one-versus-one strategy. Although several approaches based on Regions of Interest (ROIs) extraction exist, the prediction power of features has not yet investigated by comparing filter and wrapper techniques. The findings of this work suggest that (i) the IntraCranial Volume (ICV) normalization can lead to overfitting and worst the accuracy prediction of test set and (ii) the combined use of a Random Forest-based filter with a Support Vector Machines-based wrapper, improves accuracy of binary classification.

Distance from forest edge affects bee pollinators in oilseed rape fields

Wild pollinators have been shown to enhance the pollination of Brassica napus(oilseed rape) and thus increase its market value. Several studies have previously shown that pollination services are greater in crops adjoining forest patches or other seminatural habitats than in crops completely surrounded by other crops. In this study, we investigated the specific importance of forest edges in providing potential pollinators in B. napus fields in two areas in France. Bees were caught with yellow pan traps at increasing distances from both warm and cold forest edges into B. napus fields during the blooming period. A total of 4594 individual bees, representing six families and 83 taxa, were collected. We found that both bee abundance and taxa richness were negatively affected by the distance from forest edge. However, responses varied between bee groups and edge orientations. The ITD (Inter-Tegular distance) of the species, a good proxy for bee foraging range, seems to limit how far the bees can travel from the forest edge. We found a greater abundance of cuckoo bees (Nomada spp.) of Andrena spp. and Andrena spp. males at forest edges, which we assume indicate suitable nesting sites, or at least mating sites, for some abundant Andrena species and their parasites (Fig. 1). Synthesis and Applications. This study provides one of the first examples in temperate ecosystems of how forest edges may actually act as a reservoir of potential pollinators and directly benefit agricultural crops by providing nesting or mating sites for important early spring pollinators. Policymakers and land managers should take forest edges into account and encourage their protection in the agricultural matrix to promote wild bees and their pollination services.

Novel single trial movement classification based on temporal dynamics of EEG

Various complex oscillatory processes are involved in the generation of the motor command. The temporal dynamics of these processes were studied for movement detection from single trial electroencephalogram (EEG). Autocorrelation analysis was performed on the EEG signals to find robust markers of movement detection. The evolution of the autocorrelation function was characterised via the relaxation time of the autocorrelation by exponential curve fitting. It was observed that the decay constant of the exponential curve increased during movement, indicating that the autocorrelation function decays slowly during motor execution. Significant differences were observed between movement and no moment tasks. Additionally, a linear discriminant analysis (LDA) classifier was used to identify movement trials with a peak accuracy of 74%.

Classification of seed storage behaviour of 67 Amazonian tree species

Information was collated on the seed storage behaviour of 67 tree species native to the Amazon rainforest of Brazil; 38 appeared to show orthodox, 23 recalcitrant and six intermediate seed storage behaviour. A double-criteria key based on thousand-seed weight and seed moisture content at shedding to estimate likely seed storage behaviour, developed previously, showed good agreement with the above classifications. The key can aid seed storage behaviour identification considerably.

Smart-phone based electrocardiogram wavelet decomposition and neural network classification

This paper discusses ECG classification after parametrizing the ECG waveforms in the wavelet domain. The aim of the work is to develop an accurate classification algorithm that can be used to diagnose cardiac beat abnormalities detected using a mobile platform such as smart-phones. Continuous time recurrent neural network classifiers are considered for this task. Records from the European ST-T Database are decomposed in the wavelet domain using discrete wavelet transform (DWT) filter banks and the resulting DWT coefficients are filtered and used as inputs for training the neural network classifier. Advantages of the proposed methodology are the reduced memory requirement for the signals which is of relevance to mobile applications as well as an improvement in the ability of the neural network in its generalization ability due to the more parsimonious representation of the signal to its inputs.

On the application of optimal wavelet filter banks for ECG signal classification

This paper discusses ECG signal classification after parametrizing the ECG waveforms in the wavelet domain. Signal decomposition using perfect reconstruction quadrature mirror filter banks can provide a very parsimonious representation of ECG signals. In the current work, the filter parameters are adjusted by a numerical optimization algorithm in order to minimize a cost function associated to the filter cut-off sharpness. The goal consists of achieving a better compromise between frequency selectivity and time resolution at each decomposition level than standard orthogonal filter banks such as those of the Daubechies and Coiflet families. Our aim is to optimally decompose the signals in the wavelet domain so that they can be subsequently used as inputs for training to a neural network classifier.

The response of the sea ice edge to atmospheric and oceanic jet formation

The sea ice edge presents a region of many feedback processes between the atmosphere, ocean, and sea ice (Maslowski et al.). Here the authors focus on the impact of on-ice atmospheric and oceanic flows at the sea ice edge. Mesoscale jet formation due to the Coriolis effect is well understood over sharp changes in surface roughness such as coastlines (Hunt et al.). This sharp change in surface roughness is experienced by the atmosphere and ocean encountering a compacted sea ice edge. This paper presents a study of a dynamic sea ice edge responding to prescribed atmospheric and oceanic jet formation. An idealized analytical model of sea ice drift is developed and compared to a sea ice climate model [the Los Alamos Sea Ice Model (CICE)] run on an idealized domain. The response of the CICE model to jet formation is tested at various resolutions. It is found that the formation of atmospheric jets at the sea ice edge increases the wind speed parallel to the sea ice edge and results in the formation of a sea ice drift jet in agreement with an observed sea ice drift jet (Johannessen et al.). The increase in ice drift speed is dependent upon the angle between the ice edge and wind and results in up to a 40% increase in ice transport along the sea ice edge. The possibility of oceanic jet formation and the resultant effect upon the sea ice edge is less conclusive. Observations and climate model data of the polar oceans have been analyzed to show areas of likely atmospheric jet formation, with the Fram Strait being of particular interest.

Computationally efficient rule-based classification for continuous streaming data

Advances in hardware and software technologies allow to capture streaming data. The area of Data Stream Mining (DSM) is concerned with the analysis of these vast amounts of data as it is generated in real-time. Data stream classification is one of the most important DSM techniques allowing to classify previously unseen data instances. Different to traditional classifiers for static data, data stream classifiers need to adapt to concept changes (concept drift) in the stream in real-time in order to reflect the most recent concept in the data as accurately as possible. A recent addition to the data stream classifier toolbox is eRules which induces and updates a set of expressive rules that can easily be interpreted by humans. However, like most rule-based data stream classifiers, eRules exhibits a poor computational performance when confronted with continuous attributes. In this work, we propose an approach to deal with continuous data effectively and accurately in rule-based classifiers by using the Gaussian distribution as heuristic for building rule terms on continuous attributes. We show on the example of eRules that incorporating our method for continuous attributes indeed speeds up the real-time rule induction process while maintaining a similar level of accuracy compared with the original eRules classifier. We termed this new version of eRules with our approach G-eRules.

Towards a parallel computationally efficient approach to scaling up data stream classification

Advances in hardware technologies allow to capture and process data in real-time and the resulting high throughput data streams require novel data mining approaches. The research area of Data Stream Mining (DSM) is developing data mining algorithms that allow us to analyse these continuous streams of data in real-time. The creation and real-time adaption of classification models from data streams is one of the most challenging DSM tasks. Current classifiers for streaming data address this problem by using incremental learning algorithms. However, even so these algorithms are fast, they are challenged by high velocity data streams, where data instances are incoming at a fast rate. This is problematic if the applications desire that there is no or only a very little delay between changes in the patterns of the stream and absorption of these patterns by the classifier. Problems of scalability to Big Data of traditional data mining algorithms for static (non streaming) datasets have been addressed through the development of parallel classifiers. However, there is very little work on the parallelisation of data stream classification techniques. In this paper we investigate K-Nearest Neighbours (KNN) as the basis for a real-time adaptive and parallel methodology for scalable data stream classification tasks.

Classification in e-procurement

Three coupled knowledge transfer partnerships used pattern recognition techniques to produce an e-procurement system which, the National Audit Office reports, could save the National Health Service £500 m per annum. An extension to the system, GreenInsight, allows the environmental impact of procurements to be assessed and savings made. Both systems require suitable products to be discovered and equivalent products recognised, for which classification is a key component. This paper describes the innovative work done for product classification, feature selection and reducing the impact of mislabelled data.

Application of complex extreme learning machine to multiclass classification problems with high dimensionality: a THz spectra classification problem

We extend extreme learning machine (ELM) classifiers to complex Reproducing Kernel Hilbert Spaces (RKHS) where the input/output variables as well as the optimization variables are complex-valued. A new family of classifiers, called complex-valued ELM (CELM) suitable for complex-valued multiple-input–multiple-output processing is introduced. In the proposed method, the associated Lagrangian is computed using induced RKHS kernels, adopting a Wirtinger calculus approach formulated as a constrained optimization problem similarly to the conventional ELM classifier formulation. When training the CELM, the Karush–Khun–Tuker (KKT) theorem is used to solve the dual optimization problem that consists of satisfying simultaneously smallest training error as well as smallest norm of output weights criteria. The proposed formulation also addresses aspects of quaternary classification within a Clifford algebra context. For 2D complex-valued inputs, user-defined complex-coupled hyper-planes divide the classifier input space into four partitions. For 3D complex-valued inputs, the formulation generates three pairs of complex-coupled hyper-planes through orthogonal projections. The six hyper-planes then divide the 3D space into eight partitions. It is shown that the CELM problem formulation is equivalent to solving six real-valued ELM tasks, which are induced by projecting the chosen complex kernel across the different user-defined coordinate planes. A classification example of powdered samples on the basis of their terahertz spectral signatures is used to demonstrate the advantages of the CELM classifiers compared to their SVM counterparts. The proposed classifiers retain the advantages of their ELM counterparts, in that they can perform multiclass classification with lower computational complexity than SVM classifiers. Furthermore, because of their ability to perform classification tasks fast, the proposed formulations are of interest to real-time applications.

On the microphysical effects of observed cloud edge charging

Liquid layer clouds are abundant globally. Lacking strong convection, they do not become electrified by the usual thunderstorm mechanisms of collisional electrification between hydrometeors of different phases. Instead, the background global circuit current flow in fair weather is largely unaffected by the layer cloud’s presence, and, if the layer cloud is extensive horizontally, the vertical fair weather conduction current passes through the cloud. A consequence of the vertical current flow is that, at the cloud-air boundary where there is a conductivity transition and droplets form or evaporate, droplet charging occurs. Charge can affect both droplet evaporation and droplet-droplet collisions. Using new radiosonde instrumentation, the charge observed at layer cloud edges is evaluated for both these microphysical droplet processes. This shows that the charging is more likely to affect collision processes than activation, for small droplets. Enhancing the collection efficiency of small droplets modifies their evolution and propagates through the size distribution to shorten the autoconversion timescale to rain drops, and the cloud radiative properties. Because the conduction current density is influenced by both external (e.g. solar modulation of high energy particles) and internal (e.g. ENSO) factors, current flow leading to layer cloud edge charging provides a possible route for expressing solar influences on the climate system and a teleconnection mechanism for communicating internal climate variability.

Engineering the electronic bandgaps and band edge positions in carbon-substituted 2D boron nitride: a first-principles investigation

Modification of graphene to open a robust gap in its electronic spectrum is essential for its use in field effect transistors and photochemistry applications. Inspired by recent experimental success in the preparation of homogeneous alloys of graphene and boron nitride (BN), we consider here engineering the electronic structure and bandgap of C2xB1−xN1−x alloys via both compositional and configurational modification. We start from the BN end-member, which already has a large bandgap, and then show that (a) the bandgap can in principle be reduced to about 2 eV with moderate substitution of C (x < 0.25); and (b) the electronic structure of C2xB1−xN1−x can be further tuned not only with composition x, but also with the configuration adopted by C substituents in the BN matrix. Our analysis, based on accurate screened hybrid functional calculations, provides a clear understanding of the correlation found between the bandgap and the level of aggregation of C atoms: the bandgap decreases most when the C atoms are maximally isolated, and increases with aggregation of C atoms due to the formation of bonding and anti-bonding bands associated with hybridization of occupied and empty defect states. We determine the location of valence and conduction band edges relative to vacuum and discuss the implications on the potential use of 2D C2xB1−xN1−x alloys in photocatalytic applications. Finally, we assess the thermodynamic limitations on the formation of these alloys using a cluster expansion model derived from first-principles.