Parameter-free Approach Research Articles

A new data-driven method for microarray data classification

Knowledge gained through classification of microarray gene expression data is increasingly important as they are useful for phenotype classification of diseases. In this paper we propose a rule-based approach called 'Large Coverage Rule' for microarray data classification. The proposed approach is a parameter-free data-driven approach that constructs decision rule based on the expression values of a gene. A simple Rank-Based Scoring algorithm is proposed for selecting informative genes. The performance of the proposed approach is evaluated using ten publicly available gene expression data sets. From the simulation result, it is found that the proposed approach generates compact rules and produces comparatively good classification accuracy than the others. Gene ontology based biological semantics is also carried out to analyse the informative genes. Statistical analysis of test result shows that the generated rules are simple to interpret, highly comprehensible and classifies microarray data accurately.

Serum BDNF correlates with connectivity in the (pre)motor hub in the aging human brain—a resting-state fMRI pilot study

Brain-derived neurotrophic factor (BDNF) has been discussed to be involved in plasticity processes in the human brain, in particular during aging. Recently, aging and its (neurodegenerative) diseases have increasingly been conceptualized as disconnection syndromes. Here, connectivity changes in neural networks (the connectome) are suggested to be the most relevant and characteristic features for such processes or diseases. To further elucidate the impact of aging on neural networks, we investigated the interaction between plasticity processes, brain connectivity, and healthy aging by measuring levels of serum BDNF and resting-state fMRI data in 25 young (mean age 24.8 ± 2.7 (SD) years) and 23 old healthy participants (mean age, 68.6 ± 4.1 years). To identify neural hubs most essentially related to serum BDNF, we applied graph theory approaches, namely the new data-driven and parameter-free approach eigenvector centrality (EC) mapping. The analysis revealed a positive correlation between serum BDNF and EC in the premotor and motor cortex in older participants in contrast to young volunteers, where we did not detect any association. This positive relationship between serum BDNF and EC appears to be specific for older adults. Our results might indicate that the amount of physical activity and learning capacities, leading to higher BDNF levels, increases brain connectivity in (pre)motor areas in healthy aging in agreement with rodent animal studies. Pilot results have to be replicated in a larger sample including behavioral data to disentangle the cause for the relationship between BDNF levels and connectivity.

Fully Parameter-Free Calculation of Optical Spectra for Insulators, Semiconductors, and Metals from a Simple Polarization Functional.

We present a fully parameter-free density-functional approach for the accurate description of optical absorption spectra of insulators, semiconductors, and metals. We show that this can be achieved within time-dependent current-density-functional theory using a simple dynamical polarization functional. We derive this functional from physical principles that govern optical spectra. Our method is truly predictive because not a single parameter is used. In particular, we do not use an ad hoc material-dependent broadening parameter to compare theory to experiment as is usually done. Our approach is numerically efficient; the cost equals that of a calculation within the random-phase approximation.

Electric quadrupole and magnetic dipole moments of odd nuclei near the magic ones in a self-consistent approach

We present a model which describes the properties of odd-even nuclei with one nucleon more, or less, with respect to the magic number. In addition to the effects related to the unpaired nucleon, we consider those produced by the excitation of the closed shell core. By using a single particle basis generated with Hartree-Fock calculations, we describe the polarization of the doubly magic-core with Random Phase Approximation collective wave functions. In every step of the calculation, and for all the nuclei considered, we use the same finite-range nucleon-nucleon interaction. We apply our model to the evaluation of electric quadrupole and magnetic dipole moments of odd-even nuclei around oxygen, calcium, zirconium, tin and lead isotopes. Our Random Phase Approximation description of the polarization of the core improves the agreement with experimental data with respect to the predictions of the independent particle model. We compare our results with those obtained in first-order perturbation theory, with those produced by Hartree-Fock-Bogolioubov calculations and with those generated within the Landau-Migdal theory of finite Fermi systems. The results of our universal, self-consistent, and parameter free approach have the same quality of those obtained with phenomenological approaches where the various terms of the nucleon-nucleon interaction are adapted to reproduce some specific experimental data. A critical discussion on the validity of the model is presented.

Energy-Efficient Bandwidth and Power Allocation for Multi-Homing Networks

This paper investigates resource allocation for multi-homing networks where users can simultaneously transmit data to multiple radio access networks (RANs) using multiple air interfaces. We aim at optimally assigning the bandwidth and power to each user-RAN connection so as to maximize energy-efficiency of the entire network subject to user specific QoS requirements as well as the available resource budgets. First, we study the problem of resource allocation for scenarios where the connections between the users and the RANs are predefined, which naturally leads to a fractional program. To obtain the optimal solution efficiently and facilitate distributed implementations, we further equivalently transform the design problem into a convex program using a parameter-free approach and develop a decentralized algorithm based on the alternating direction method of multipliers. Second, we investigate the problem of joint link selection and resource allocation for energy-efficiency maximization. The problem is cast as a mixed integer nonlinear convex program for which we particularize the branch and bound method to find an optimal solution. Then, two suboptimal low-complexity designs are proposed: the first one is based on greedy algorithm, which gradually maximizes the virtual link energy-efficiency; the other one is based on the concept of sparsity-inducing norm. Simulation results are presented to demonstrate the potential gains in terms of energy-efficiency of the proposed methods.

Studies of cross sections for collisions of electrons from hydride molecules:NH3andPH3

A parameter-free spherical complex optical potential approach in the fixed-nuclei approximation is applied here to calculate elastic differential, integral, and momentum-transfer cross sections as well as total (elastic plus inelastic) cross sections for the scattering of electrons from $\mathrm{N}{\mathrm{H}}_{3}$ and $\mathrm{P}{\mathrm{H}}_{3}$ molecules (i.e., nonmetallic hydrides) in the electron energy range of 0.1--100 eV. The projectile-target interaction is represented by a sum of a real and an imaginary potential. Present calculated results are compared with the available calculations and the experimental measurements. The quantitative features of the scattering parameters (such as Ramsauer-Townsend minimum and shape resonance structure), as observed in experiments, are well reproduced in the present study.

Parameter-free, weak imposition of Dirichlet boundary conditions and coupling of trimmed and non-conforming patches

We present a parameter-free domain sewing approach for low-order as well as high-order finite elements. Its final form contains only primal unknowns; that is, the approach does not introduce additional unknowns at the interface. Additionally, it does not involve problem-dependent parameters, which require an estimation. The presented approach is symmetry preserving; that is, the resulting discrete form of an elliptic equation will remain symmetric and positive definite. It preserves the order of the underlying discretization, and we demonstrate high-order accuracy for problems of non-matching discretizations concerning the mesh size h as well as the polynomial degree of the order of discretization p. We also demonstrate how the method may be used to model material interfaces, which may be curved and for which the interface does not coincide with the underlying mesh. This novel approach is presented in the context of the p-version and B-spline version of the finite cell method, an embedded domain method of high order, and compared with more classical methods such as the penalty method or Nitsche's method.

Ab initiofriction forces on the nanoscale: A density functional theory study of fcc Cu(111)

While there are a number of models that tackle the problem of calculating friction forces on the atomic level, providing a completely parameter-free approach remains a challenge. Here we present a quasi-static model to obtain an approximation to the nanofrictional response of dry, wearless systems based on quantum mechanical all-electron calculations. We propose a mechanism to allow dissipative sliding, which relies on atomic relaxations. We define two different ways of calculating the mean nanofriction force, both leading to an exponential friction-versus-load behavior for all sliding directions. Since our approach does not impose any limits on lengths and directions of the sliding paths, we investigate arbitrary sliding directions for an fcc Cu(111) interface and detect two periodic paths which form the upper and lower bound of nanofriction. For long aperiodic paths the friction force convergences to a value in between these limits. For low loads we retrieve the Derjaguin generalization of Amontons-Coulomb kinetic friction law which appears to be valid all the way down to the nanoscale. We observe a non-vanishing Derjaguin-offset even for atomically flat surfaces in dry contact.

Bayesian CP Factorization of Incomplete Tensors with Automatic Rank Determination.

CANDECOMP/PARAFAC (CP) tensor factorization of incomplete data is a powerful technique for tensor completion through explicitly capturing the multilinear latent factors. The existing CP algorithms require the tensor rank to be manually specified, however, the determination of tensor rank remains a challenging problem especially for CP rank . In addition, existing approaches do not take into account uncertainty information of latent factors, as well as missing entries. To address these issues, we formulate CP factorization using a hierarchical probabilistic model and employ a fully Bayesian treatment by incorporating a sparsity-inducing prior over multiple latent factors and the appropriate hyperpriors over all hyperparameters, resulting in automatic rank determination. To learn the model, we develop an efficient deterministic Bayesian inference algorithm, which scales linearly with data size. Our method is characterized as a tuning parameter-free approach, which can effectively infer underlying multilinear factors with a low-rank constraint, while also providing predictive distributions over missing entries. Extensive simulations on synthetic data illustrate the intrinsic capability of our method to recover the ground-truth of CP rank and prevent the overfitting problem, even when a large amount of entries are missing. Moreover, the results from real-world applications, including image inpainting and facial image synthesis, demonstrate that our method outperforms state-of-the-art approaches for both tensor factorization and tensor completion in terms of predictive performance.

A density and connectivity based decision rule for pattern classification

In this paper we propose a novel neighborhood classifier, Surrounding Influence Region (SIR) decision rule. Traditional Nearest Neighbor (NN) classifier is a distance-based method, and it classifies a sample using a predefined number of neighbors. In this study neighbors of a sample are determined using not only the distance, but also the connectivity and density information. One of the well-known proximity graphs, Gabriel Graph, is used for this purpose. The neighborhood is unique for each sample. SIR decision rule is a parameter-free approach. Our experiments with artificial and real data sets show that the performance of the SIR decision rule is superior to the k-NN and Gabriel Graph neighbor (GGN) classifiers in most of the data sets.

Ab Initio Treatment of Disorder Effects in Amorphous Organic Materials: Toward Parameter Free Materials Simulation.

Disordered organic materials have a wide range of interesting applications, such as organic light emitting diodes, organic photovoltaics, and thin film electronics. To model electronic transport through such materials it is essential to describe the energy distribution of the available electronic states of the carriers in the material. Here, we present a self-consistent, linear-scaling first-principles approach to model environmental effects on the electronic properties of disordered molecular systems. We apply our parameter free approach to calculate the energy disorder distribution of localized charge states in a full polaron model for two widely used benchmark-systems (tris(8-hydroxyquinolinato)aluminum (Alq3) and N,N'-bis(1-naphthyl)-N,N'-diphenyl-1,1'-biphenyl-4,4'-diamine (α-NPD)) and accurately reproduce the experimental charge carrier mobility over a range of 4 orders of magnitude. The method can be generalized to determine electronic and optical properties of more complex systems, e.g. guest-host morphologies, organic-organic interfaces, and thus offers the potential to significantly contribute to de novo materials design.

A shrinking hypersphere PSO for engineering optimisation problems

Many real-world and engineering design problems can be formulated as constrained optimisation problems (COPs). Swarm intelligence techniques are a good approach to solve COPs. In this paper an efficient shrinking hypersphere-based particle swarm optimisation (SHPSO) algorithm is proposed for constrained optimisation. The proposed SHPSO is designed in such a way that the movement of the particle is set to move under the influence of shrinking hyperspheres. A parameter-free approach is used to handle the constraints. The performance of the SHPSO is compared against the state-of-the-art algorithms for a set of 24 benchmark problems. An exhaustive comparison of the results is provided statistically as well as graphically. Moreover three engineering design problems namely welded beam design, compressed string design and pressure vessel design problems are solved using SHPSO and the results are compared with the state-of-the-art algorithms.

A parameter-free approach for one-class classification using binary decision diagrams

Parameter tuning is essential in classification problems to achieve a high performance, but it is very hard when it comes to the one-class classification problem. In this paper, we propose a novel one-class classifier whose parameter can be tuned automatically. The proposed classifier can deal with non-linearly distributed data and is robust to noise in training data sets. Moreover, the proposed classifier can be learnt efficiently in the case that a training data set is large, because the computational complexity is approximately linear with respect to the number of training data. In the proposed method, the region of a training data set is expressed as a Boolean formula that is constructed by using a binary decision diagram. Then the region is efficiently over-approximated through the direct manipulation of the binary decision diagram. The parameter of the over-approximation can be tuned automatically based on the minimum description length principle. Experimental results show that the proposed method works very well with synthetic data and some realistic data.

Study of nucleonic matter with a consistent two- and three-body perturbative chiral interaction

We calculate perturbatively the energy per nucleon in infinite nuclear matter with a chiral N3LO (next-to-next-to-next-to-leading order) two-body potential plus a N2LO three- body force (3BF). The 3BF low-energy constants which cannot be constrained by two-body observables are chosen such as to reproduce the A = 3 binding energies and the triton Gamow- Teller matrix element. This enables to study the nuclear matter equation of state in a parameterfree approach.

Docking covalent inhibitors: a parameter free approach to pose prediction and scoring.

Although many popular docking programs include a facility to account for covalent ligands, large-scale systematic docking validation studies of covalent inhibitors have been sparse. In this paper, we present the development and validation of a novel approach for docking and scoring covalent inhibitors, which consists of conventional noncovalent docking, heuristic formation of the covalent attachment point, and structural refinement of the protein-ligand complex. This approach combines the strengths of the docking program Glide and the protein structure modeling program Prime and does not require any parameter fitting for the study of additional covalent reaction types. We first test this method by predicting the native binding geometry of 38 covalently bound complexes. The average RMSD of the predicted poses is 1.52 Å, and 76% of test set inhibitors have an RMSD of less than 2.0 Å. In addition, the apparent affinity score constructed herein is tested on a virtual screening study and the characterization of the SAR properties of two different series of congeneric compounds with satisfactory success.

On the discrete variant of the traction method in parameter-free shape optimization

One of the major challenges in the parameter-free approach to computational shape optimization is the avoidance of oscillating (i.e. non-smooth) boundaries in the optimal design trials. In order to achieve this, we investigate a method for regularization that corresponds to the discrete variant of the so-called traction method. In this approach, the design updates are generated in terms of a displacement field, which is obtained as the solution to an auxiliary boundary value problem that is defined on the actual design domain. The main idea herein is to apply fictitious nodal forces corresponding to the discrete sensitivity of the objective function. We propose an algorithm in which constraint functions will be taken into account by using an augmented Lagrangian formulation and a step-length control ensures a sufficient decrease condition in terms of the objective function within each iteration. We examine the benefits of the proposed regularization method on the basis of some numerical examples in comparison to an unregularized steepest-descent algorithm.

Serum Neuron-Specific Enolase Is Related to Cerebellar Connectivity: A Resting-State Functional Magnetic Resonance Imaging Pilot Study.

Neuron-specific enolase (NSE) has been suggested as a prognostic biomarker for neuronal alterations resulting from conditions such as traumatic brain injury (TBI), neurodegenerative disease, or cardiac arrest. To validate serum NSE (sNSE) as a brain-specific biomarker, we related it to functional brain imaging data in 38 healthy adults to create a physiological framework for future studies in neuropsychiatric diseases. sNSE was measured by monoclonal two-site immunoluminometric assays, and functional connectivity was investigated with resting-state functional magnetic resonance imaging (rfMRI). To identify neural hubs most essentially related to sNSE, we applied graph theory approaches, namely, the new data-driven and parameter-free approach, eigenvector centrality mapping. sNSE and eigenvector centrality were negatively correlated in the female cerebellum, without any effects in male subjects. In cerebellar cortex, NSE expression was significantly higher than whole-brain expression as investigated in the whole brain and whole genome-wide atlas of the Allen Institute for Brain Sciences (Seattle, WA). Our study shows a specific linkage between the neuronal marker protein, sNSE, and cerebellar connectivity as measured with rfMRI in the female human brain, although this finding shall be proven in future studies including more subjects. Results suggest that the inclusion of sNSE in the analysis of imaging data is a useful approach to obtain more-specific information on the neuronal mechanisms that underlie functional connectivity at rest. Establishing such a baseline resting-state pattern that is tied to a neuronal serum marker opens new perspectives in the characterization of neuropsychiatric disorders as disconnective syndromes or nexopathies, in particular, resulting from TBI, neurodegenerative disease, or cardiac arrest, in the future.

Polymer-mediated tunneling transport between carbon nanotubes in nanocomposites.

Electron transport in nanocomposites has attracted a good deal of attention for some time now; furthermore, the ability to control its characteristics is a necessary step in the design of multifunctional materials. When conductive nanostructures (for example carbon nanotubes) are inserted in a non-conductive matrix, electron transport below the percolation threshold is dominated by tunneling and thus the conductive characteristics of the composite depends heavily on the characteristics of the tunneling currents between nanoinserts. A parameter-free approach to study tunneling transport between carbon nanotubes across a polymer matrix is presented. The calculation is done with a combination of Density Functional Theory and Green functions (an approach heavily used in molecular electronics) which is shown here to be effective in this non-resonant transport condition. The results show that the method can effectively capture the effect of a dielectric layer in tunneling transport. The current is found to exponentially decrease with the size of the gap for both vacuum and polymer, and that the polymer layer lowers the tunneling barrier enhancing tunneling conduction. For a polyacrylonitrile matrix, a four-fold decrease in the tunneling constant, compared to tunneling in vacuum, is observed, a result that is consistent with available information. The method is very versatile as any DFT functional (or any other quantum mechanics method) can be used and thus the most accurate method for each particular system can be chosen. Furthermore as more methods become available, the calculations can be revised and improved. This approach can be used to design functional materials for fine-tunning the tunneling transport, for instance, the effect of modifying the nanoinsert-matrix interface (for example, by adding functional groups to carbon nanotubes) can be captured and the comparative performance of each interface predicted by simulation.

Nuclear-matter equation of state with consistent two- and three-body perturbative chiral interactions

We compute the energy per particle of infinite symmetric nuclear matter from chiral N$^{3}\mathrm{LO}$ (next-to-next-to-next-to-leading order) two-body potentials plus N$^{2}\mathrm{LO}$ three-body forces. The low-energy constants of the chiral three-nucleon force that cannot be constrained by two-body observables are fitted to reproduce the triton binding energy and the $^{3}\mathrm{H}$-$^{3}\mathrm{He}$ Gamow-Teller transition matrix element. In this way, the saturation properties of nuclear matter are reproduced in a parameter-free approach. The equation of state is computed up to third order in many-body perturbation theory, with special emphasis on the role of the third-order particle-hole diagram. The dependence of these results on the cutoff scale and regulator function is studied. We find that the inclusion of three-nucleon forces consistent with the applied two-nucleon interaction leads to a reduced dependence on the choice of the regulator only for lower values of the cutoff.

On the spatial representativeness of temporal dynamics at European weather stations

The prospects of validating areal data from climate models by site observations depend critically on the spatial representativeness of the sites. This paper introduces a simple parameter-free approach to quantify the spatial representativeness of single stations with an application to time series of daily near surface air temperature and precipitation at European weather stations. Complementing classical methods such as spatial auto-correlation and variogram, our approach provides a well defined area around each station for which the station is representative, based on a similarity threshold but otherwise without any limiting assumptions about distribution or spatial stationarity of the data. This area is interpreted as the station's ‘inverse footprint’ and its areal extent provides a measure of representativeness. We find a generally higher representativeness for temperature compared to precipitation, but also a strong seasonal dependence. For instance, temperature representativeness in boreal winter is related to the influence of circulation with large ‘inverse footprints’ over Central and Eastern Europe and Scandinavia. Representativeness in summer exhibits similar patterns but is lower. Precipitation representativeness displays related patterns of representativeness and circulation control in winter, but vanishes in summer, probably due to the small-scale characteristics of convective precipitation. Precipitation representativeness is strikingly high around the Mediterranean, which is a consequence of the large numbers of synchronous dry days in this region. The physical plausibility of the results underlines the applicability of our approach. Although not investigated here, the ‘inverse footprint’ also provides information on the directional dependence of representativeness at the station level.