Neighbor Correlation Research Articles

Oversampling multi-label data based on natural neighbor and label correlation

In multi-label learning, addressing the class imbalance issue is of paramount importance. Oversampling methods are preferred as they offer a more general solution independent of the model choice, i.e., they alleviate the imbalance of datasets by augmenting instances in the pre-processing step. Existing neighbor-based oversampling methods employ an empirical number of neighbors (k=5) to identify the local region for new instances creation. However, a single fixed k value cannot fit all labels, because every label usually has its own distinct distribution and complexity. Furthermore, the label assignment for synthetic instances usually depends on the statistics of individual labels within the corresponding neighborhood, ignoring the informative correlation among labels. To overcome these limitations, we propose an oversampling method called Multi-Label Oversampling with Natural neighbor and label Correlation (MLONC). Our approach offers three main advantages: (1) the adaptive number of neighbors for each label related to the data complexity is obtained via natural neighbor detection; (2) it encourages generating more instances proximate to the decision boundary of highly imbalanced labels, and diminishes the impact of outliers; (3) exploitation of label correlation in label assignment enhances the quality of the synthetic instances. Experimental results demonstrate the effectiveness of MLONC under various base classifiers.

A fusiform network of indoor scene classification with the stylized semantic description for service-robot applications

Scene perception technology helps robots identify the target areas that people refer to so that it contributes to human–robot interaction, semantic navigation and other related tasks. Currently, pure semantic-feature-based methods are insufficient to fully describe diverse indoor scenes, resulting in a high confusion rate and performance inconsistency on each class. To overcome these problems, the style information is compounded with the common semantic feature to form a more elaborate description of a scene, and the corresponding network is proposed. First, the convolutional network is adopted to extract the base feature. Then, the high-level feature maps are taken out and reasonably divided into overlapped units to reserve a more appropriate neighbour correlation. Next, two branches are proposed to acquire the style and the semantic information respectively. In the style branch, the gram matrix is applied to each unit. In the semantic branch, the units of high-level feature maps are directly used. In both branches, batch normalization and vector embedding techniques are applied to the flattened elements of the unit sets to unify the feature strength and introduce compression representation. Then, the two compressed representations are combined as a compound expression to describe a scene. A multi-head self-attention structure is introduced to correlate and reinforce the multi-divided information and further form a dominant and refined stylized semantic description. Finally, scene classification is implemented by a multi-layer fully connected network. The experiment adopts the paradigm of once learning and cross-environment inference, which is closer to practical applications. The proposed method performs the best compared with several popular methods in the field of robotics, and especially, it has the smallest classification bias. In addition, the necessity of the main components of our method is evaluated, and the semantic explanation is also presented.

Effect of neighbouring molecules on ground-state properties of many-body polar linear rotor systems

A path integral ground state approach has been used to estimate the ground-state energy and structural properties of hydrogen fluoride molecules pinned to a one-dimensional lattice. In the simulations, the molecules are assumed to be rigid, and only the continuous rotational degrees of freedom are considered. The constituents of a many-body system interact through the dipole-dipole interaction because the molecules have a permanent dipole moment. The workability of our approach has been demonstrated by estimating the ground-state energy, order parameter and nearest neighbour correlation for the systems of 2 and 3 HF molecules using quantum Monte Carlo simulations based on the path integral ground state methodology. The results agree satisfactorily with those obtained from exact Hamiltonian matrix diagonalisation. In addition, the effect of neighbours on the ground state properties has been investigated for larger systems. The converged ground-state energy per neighbour as a function of inter-nuclear separation is considered to be the equation of state for the system. The chemical potential of the rotor chains also supports the convergence test.

Robust Point Cloud Segmentation with Noisy Annotations.

Point cloud segmentation is a fundamental task in 3D. Despite recent progress on point cloud segmentation with the power of deep networks, current learning methods based on the clean label assumptions may fail with noisy labels. Yet, class labels are often mislabeled at both instance-level and boundary-level in real-world datasets. In this work, we take the lead in solving the instance-level label noise by proposing a Point Noise-Adaptive Learning (PNAL) framework. Compared to noise-robust methods on image tasks, our framework is noise-rate blind, to cope with the spatially variant noise rate specific to point clouds. Specifically, we propose a point-wise confidence selection to obtain reliable labels from the historical predictions of each point. A cluster-wise label correction is proposed with a voting strategy to generate the best possible label by considering the neighbor correlations. To handle boundary-level label noise, we also propose a variant "PNAL-boundary" with a progressive boundary label cleaning strategy. Extensive experiments demonstrate its effectiveness on both synthetic and real-world noisy datasets. Even with 60% symmetric noise and high-level boundary noise, our framework significantly outperforms its baselines, and is comparable to the upper bound trained on completely clean data. Moreover, we cleaned the popular real-world dataset ScanNetV2 for rigorous experiment. Our code and data is available at https://github.com/pleaseconnectwifi/PNAL.

Belief propagation on networks with cliques and chordless cycles.

It is well known that tree-based theories can describe the properties of undirected clustered networks with extremely accurate results [S. Melnik etal., Phys. Rev. E 83, 036112 (2011)10.1103/PhysRevE.83.036112]. It is reasonable to suggest that a motif-based theory would be superior to a tree one, since additional neighbor correlations are encapsulated in the motif structure. In this paper, we examine bond percolation on random and real world networks using belief propagation in conjunction with edge-disjoint motif covers. We derive exact message passing expressions for cliques and chordless cycles of finite size. Our theoretical model gives good agreement with Monte Carlo simulation and offers a simple, yet substantial improvement on traditional message passing, showing that this approach is suitable to study the properties of random and empirical networks.

BigNeuron: a resource to benchmark and predict performance of algorithms for automated tracing of neurons in light microscopy datasets.

BigNeuron is an open community bench-testing platform with the goal of setting open standards for accurate and fast automatic neuron tracing. We gathered a diverse set of image volumes across several species that is representative of the data obtained in many neuroscience laboratories interested in neuron tracing. Here, we report generated gold standard manual annotations for a subset of the available imaging datasets and quantified tracing quality for 35 automatic tracing algorithms. The goal of generating such a hand-curated diverse dataset is to advance the development of tracing algorithms and enable generalizable benchmarking. Together with image quality features, we pooled the data in an interactive web application that enables users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and tracing data, and benchmarking of automatic tracing algorithms in user-defined data subsets. The image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. We observed that diverse algorithms can provide complementary information to obtain accurate results and developed a method to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms in noisy datasets. However, specific algorithms may outperform the consensus tree strategy in specific imaging conditions. Finally, to aid users in predicting the most accurate automatic tracing results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic tracings.

A novel prediction method for protein DNA-binding residues based on neighboring residue correlations

Accurately identifying the protein DNA-binding residues is important for understanding the protein–DNA recognition mechanism and protein function annotation. Many computational methods have been proposed to predict protein–DNA binding residues using protein sequence information; however, for severe imbalanced data like the protein–DNA binding dataset, the under-sampling technique which is applied by most previous methods cannot achieve satisfactory performance. In this study, an adjustment algorithm is proposed to offset the biased prediction results from the classifier. The proposed adjustment algorithm uses the binding probability between the target residue and its neighboring residues to identify more true binding residues which are wrongly predicted as non-binding. The proposed prediction method with adjustment algorithm achieves an area under the ROC curve (AUC) of 0.926 and 0.866 on two benchmark datasets and 0.882 on the independent testing set, which demonstrates that the proposed method can efficiently predict specific residues for protein–DNA interactions.

Categorical Neighbour Correlation Coefficient (CnCor) for Detecting Relationships between Categorical Variables

Categorical data is common and, however, special in that its possible values exist only on a nominal scale so that many statistical operations such as mean, variance, and covariance become not applicable. Following the basic idea of the neighbour correlation coefficient (nCor), in this study, we propose a new measure named the categorical nCor (CnCor) to examine the association between categorical variables through using indicator functions to reform the distance metric and product-moment correlation coefficient. The proposed measure is easy to compute, and enables a direct test of statistical dependence without the need of converting the qualitative variables to quantitative ones. Compare to previous approaches, it is much more robust and effective in dealing with multi-categorical target variables especially when highly nonlinear relationships occurs in the multivariate case. We also applied the CnCor to implementing feature selection by the scheme of backward elimination. Finally, extensive experiments performed on both synthetic and real-world datasets are conducted to demonstrate the outstanding performance of the proposed methods, and draw comparisons with state-of-the-art association measures and feature selection algorithms.

Implementation of the lattice model in the coexistence of species and its potential consequences on environment

The population dynamics of a system of two competing species have been investigated in the mean-field and lattice approximation. The two species are denoted by A and B. Each site of the square lattice is either occupied by an individual or vacant. The two species complete for vacant sites to reproduce. There is a reproduction only to the nearest neighbours. We consider the invasion of a rare species into a population composed of a resident species based on a pair – approximation method in which the dynamics of both average densities and nearest neighbour correlations are considered. The results are then compared with those obtained by the mean-field approximation. Whenspecies B contain intraspecific interaction term, invasion of the rare species A into resident species B becomes easier in lattice structured populations. But the rare species B invading species A is difficult in lattice models in comparison to mean-field approximation. The overall coexistence of species is enhanced in lattice models. These results were verified by simulation on a square lattice although the range of the enhancement of the species coexistence is reduced. This calls for the attention that pair-approximation is an oversimplification of the real situation.

Adaptive encoding based lossless data hiding method for VQ compressed images using tabu search

A novel joint image coding and reversible data hiding method for vector quantization (VQ) compressed images is proposed in this paper. Since the original VQ indices often exhibit uncorrelated, a rearrangement of them by considering their correlations may benefit the prediction performance so as to reduce the required bitrate. In this study, the tabu search algorithm is employed to rearrange codewords by fully exploiting their neighboring correlations, yielding a moepressible rearranged indices. By combining more highly-correlated indices of a to-be-predicted index into prediction, the improved linear regression method is then applied to achieve a sharper prediction-error histogram and less required additional information. After prediction, an adaptive run-length encoding method is presented to encode prediction errors, thereby eliminating unnecessary indicators. Experimental results demonstrate that the proposed method effectively reduces the bitrate of the compressed image while providing a comparable hiding capacity.

Intermolecular correlations of liquid and glassy CS2 studied by synchrotron radiation x-ray diffraction.

How is the orientation of molecular liquids ordered on cooling? What are the basic structures of molecular glasses, e.g., close to the crystalline structure or some special structures such as icosahedral cluster? These are long-standing questions in liquid and glass physics. We have constructed a novel cryostat to prepare simple molecular glasses by vapor deposition and performed in situ synchrotron radiation x-ray diffraction experiments. The glassy state of a simple molecule CS2, which cannot be vitrified by normal liquid quenching, was successfully prepared with this instrument, and its diffraction data were collected in a wide Q-range of 0.16-25.7 Å-1 with a high-energy diffractometer at BL04B2, SPring-8. The diffraction data of liquid CS2 were also recorded in a wide temperature range of 160-300K. These diffraction data were analyzed with molecular dynamics simulations and reverse Monte Carlo modelings to investigate orientational correlation. From the obtained 3D structure models, the orientational correlation between neighboring CS2 molecules was investigated quantitatively as a function of temperature. At room temperature, the parallel and T-shaped arrangements are preferred for the nearest neighbor correlation. On cooling, these arrangements are developed gradually, and its rate became prominent below the melting temperature (162K). In the glassy state, the slipped-parallel arrangement is dominant as well as the T-shaped arrangement. Both arrangements appear in the CS2 crystal, indicating that the structure of glassy CS2 is close to that of crystalline CS2.

Magnetic polaronic and bipolaronic excitons in Mn(II) doped (TDMP)PbBr4 and their high emission

If we introduce transition metal ions with spins into an organic-inorganic halide perovskite compounds, interesting spin dependent optical properties may be seen in this system. This paper reports an organic-inorganic hybrid manganese ion doped lead halide crystal incorporated by TDMP (trans-2,5-dimethylpiperazinium), it gives not only the STE1 (polaronic exciton), usually brought up by the enhanced electron-phonon coupling in the amine incorporated compound, but also give a novel bipolaronic exciton (STE2) for the paired charge state formation due to the incorporation of bi-amine cationic molecules, a excitation by the polaronic molecule due to enhanced electron-phonon coupling and neighboring correlation out of the Mn ion ferromagnetic spin coupling in the lattice at slightly high doping. Such magnetic bipolaron state could often be accompanied by several multiphonon Raman modes and ferromagnetism. It is even interesting that this state can give super-high red emission with PLQY at ~92.5% at room temperature. Moreover, the emission color and intensity could also be modulated by the Mn doping concentration, temperature, and laser excitation power. The low doping products can give green emission out of single Mn d-d transition at very low temperature or high laser excitation power. The intermediately doped products can give white emission out of STE1 and STE2. The highly doping products give red emission out of STE2 (bipolaron states) at room temperature or low laser power. These interesting phenomena reflected the microscopic interactions between spins, carriers and phonons in the products, which can be found applications in the future fields of photonics and spintronics.

Underscreening and hidden ion structures in large scale simulations of concentrated electrolytes.

The electrostatic screening length predicted by Debye-Hückel theory decreases with increasing ionic strength, but recent experiments have found that the screening length can instead increase in concentrated electrolytes. This phenomenon, referred to as underscreening, is believed to result from ion-ion correlations and short-range forces such as excluded volume interactions among ions. We use Brownian Dynamics to simulate a version of the Restrictive Primitive Model for electrolytes over a wide range of ion concentrations, ionic strengths, and ion excluded volume radii for binary electrolytes. We measure the decay of the charge-charge correlation among ions in the bulk and compare it against scaling trends found experimentally and determined in certain weak coupling theories of ion-ion correlation. Moreover, we find that additional large scale ion structures emerge at high concentrations. In this regime, the frequency of oscillations computed from the charge-charge correlation function is not dominated by electrostatic interactions but rather by excluded volume interactions and with oscillation periods on the order of the ion diameter. We also find the nearest neighbor correlation of ions sharing the same charge transitions from negative at small concentrations to positive at high concentrations, representing the formation of small, like-charge ion clusters. We conclude that the increase in local charge density due to the formation of these clusters and the topological constraints of macroscopic charged surfaces can help explain the degree of underscreening observed experimentally.

Inferring interactions of time-delayed dynamic networks by random state variable resetting

Time delays exist widely in real systems, and time-delayed interactions can result in abundant dynamic behaviors and functions in dynamic networks. Inferring the time delays and interactions is challenging due to systematic nonlinearity, noises, a lack of information, and so on. Recently, Shi et al. proposed a random state variable resetting method to detect the interactions in a continuous-time dynamic network. By arbitrarily resetting the state variable of a driving node, the equivalent coupling functions of the driving node to any response node in the network can be reconstructed. In this paper, we introduce this method in time-delayed dynamic networks. To infer actual time delays, the nearest neighbor correlation (NNC) function for a given time delay is defined. The significant increments of NNC originate from the delayed effect. Based on the increments, the time delays can be reconstructed and the reconstruction errors depend on the sampling time interval. After time delays are accurately identified, the equivalent coupling functions can also be reconstructed. The numerical results have fully verified the validity of the theoretical analysis.

The 2D non self-dual Ising lattices: An exact renormalization group treatment

In this work, an exact renormalization group treatment of honeycomb lattice leading to an exact relation between the coupling strengths of the honeycomb and the triangular lattices is presented. Using the honeycomb and the triangular duality relation, the critical coupling values of honeycomb and triangular lattices are calculated exactly by the simultaneous solution of the renormalized relation and the duality relation, without using the so-called star-triangular transformation. Apparently, the obtained coupling relation is unique. It not only takes place the role of the star triangular relation, but it is also the only exact relation obtained from renormalization group theory other than the 1D Ising chain. An exact pair correlation function expression relating the nearest neighbors and the next nearest neighbor correlation functions are also obtained for the honeycomb lattice. Utilizing this correlation relation, an exact expression of the correlation length of the honeycomb lattice is calculated analytically for the coupling constant values less than the critical value in the realm of the scaling theory. The critical exponents [Formula: see text] and [Formula: see text] are also calculated as [Formula: see text] and [Formula: see text].

CloudLSTM: A Recurrent Neural Model for Spatiotemporal Point-cloud Stream Forecasting

This paper introduces CloudLSTM, a new branch of recurrent neural models tailored to forecasting over data streams generated by geospatial point-cloud sources. We design a Dynamic Point-cloud Convolution (DConv) operator as the core component of CloudLSTMs, which performs convolution directly over point-clouds and extracts local spatial features from sets of neighboring points that surround different elements of the input. This operator maintains the permutation invariance of sequence-to-sequence learning frameworks, while representing neighboring correlations at each time step -- an important aspect in spatiotemporal predictive learning. The DConv operator resolves the grid-structural data requirements of existing spatiotemporal forecasting models and can be easily plugged into traditional LSTM architectures with sequence-to-sequence learning and attention mechanisms. We apply our proposed architecture to two representative, practical use cases that involve point-cloud streams, i.e. mobile service traffic forecasting and air quality indicator forecasting. Our results, obtained with real-world datasets collected in diverse scenarios for each use case, show that CloudLSTM delivers accurate long-term predictions, outperforming a variety of competitor neural network models.

Phase-Aberration Correction for HIFU Therapy Using a Multielement Array and Backscattering of Nonlinear Pulses.

Phase aberrations induced by heterogeneities in body wall tissues introduce a shift and broadening of the high-intensity focused ultrasound (HIFU) focus, associated with decreased focal intensity. This effect is particularly detrimental for HIFU therapies that rely on shock front formation at the focus, such as boiling histotripsy (BH). In this article, an aberration correction method based on the backscattering of nonlinear ultrasound pulses from the focus is proposed and evaluated in tissue-mimicking phantoms. A custom BH system comprising a 1.5-MHz 256-element array connected to a Verasonics V1 engine was used as a pulse/echo probe. Pulse inversion imaging was implemented to visualize the second harmonic of the backscattered signal from the focus inside a phantom when propagating through an aberrating layer. Phase correction for each array element was derived from an aberration-correction method for ultrasound imaging that combines both the beamsum and the nearest neighbor correlation method and adapted it to the unique configuration of the array. The results were confirmed by replacing the target tissue with a fiber-optic hydrophone. Comparing the shock amplitude before and after phase-aberration correction showed that the majority of losses due to tissue heterogeneity were compensated, enabling fully developed shocks to be generated while focusing through aberrating layers. The feasibility of using a HIFU phased-array transducer as a pulse-echo probe in harmonic imaging mode to correct for phase aberrations was demonstrated.

Semi-Supervised Auto-Encoder Graph Network for Diabetic Retinopathy Grading

Diabetic Retinopathy (DR) causes quite a few blindness worldwide, which can be refrained by the timely diagnosis on retinal images. Recently, researches on deep learning-based retinal image classification have accelerated outstanding improvements in DR grading task. However, existing DR grading works are mostly limited to a supervised manner. They require accurately annotated data labeled by professional experts, and the annotating work is very laborious and time-consuming. We propose a Semi-supervised Auto-encoder Graph Network (SAGN) for the challenging DR diagnosis to relax this constraint. Precisely, SAGN consists of three major modules: auto-encoder feature learning, neighbor correlation mining, and graph representation. Firstly, our model learns to extract representations from retinal images and reconstruct them as close to original inputs as possible. Then neighbor correlations among labeled and unlabeled samples are established by their similarities, calculated by the radial basis function. Finally, we operate Graph Convolutional Neural Network (GCN) to grade retinal samples from extracted features and their correlations. To evaluate the performance of SAGN, we conduct sufficient comparative experiments on APTOS 2019 dataset, trained from EyePACS. Results demonstrate that our SAGN model can achieve comparable performance with limited labeled retinal images with the help of large amounts of unlabeled data.

The Resolution in X-ray Crystallography and Single-Particle Cryogenic Electron Microscopy

X-ray crystallography and single-particle analysis cryogenic electron microscopy are essential techniques for uncovering the three-dimensional structures of biological macromolecules. Both techniques rely on the Fourier transform to calculate experimental maps. However, one of the crucial parameters, resolution, is rather broadly defined. Here, the methods to determine the resolution in X-ray crystallography and single-particle analysis are summarized. In X-ray crystallography, it is becoming increasingly more common to include reflections discarded previously by traditionally used standards, allowing for the inclusion of incomplete and anisotropic reflections into the refinement process. In general, the resolution is the smallest lattice spacing given by Bragg’s law for a particular set of X-ray diffraction intensities; however, typically the resolution is truncated by the user during the data processing based on certain parameters and later it is used during refinement. However, at which resolution to perform such a truncation is not always clear and this makes it very confusing for the novices entering the structural biology field. Furthermore, it is argued that the effective resolution should be also reported as it is a more descriptive measure accounting for anisotropy and incompleteness of the data. In single particle cryo-EM, the situation is not much better, as multiple ways exist to determine the resolution, such as Fourier shell correlation, spectral signal-to-noise ratio and the Fourier neighbor correlation. The most widely accepted is the Fourier shell correlation using a threshold of 0.143 to define the resolution (so-called “gold-standard”), although it is still debated whether this is the correct threshold. Besides, the resolution obtained from the Fourier shell correlation is an estimate of varying resolution across the density map. In reality, the interpretability of the map is more important than the numerical value of the resolution.

Systematically Exploring Associations among Multivariate Data

Detecting relationships among multivariate data is often of great importance in the analysis of high-dimensional data sets, and has received growing attention for decades from both academic and industrial fields. In this study, we propose a statistical tool named the neighbor correlation coefficient (nCor), which is based on a new idea that measures the local continuity of the reordered data points to quantify the strength of the global association between variables. With sufficient sample size, the new method is able to capture a wide range of functional relationship, whether it is linear or nonlinear, bivariate or multivariate, main effect or interaction. The score of nCor roughly approximates the coefficient of determination (R2) of the data which implies the proportion of variance in one variable that is predictable from one or more other variables. On this basis, three nCor based statistics are also proposed here to further characterize the intra and inter structures of the associations from the aspects of nonlinearity, interaction effect, and variable redundancy. The mechanisms of these measures are proved in theory and demonstrated with numerical analyses.