Symmetric Templates Research Articles

MV-SHIF: Multi-view symmetric hypothesis inference fusion network for emotion-cause pair extraction in documents

Emotion-cause pair extraction (ECPE) is a challenging task that aims to automatically identify pairs of emotions and their causes from documents. The difficulty of ECPE lies in distinguishing valid emotion-cause pairs from many irrelevant ones. Most previous methods have primarily focused on utilizing multi-task learning to extract semantic information solely from documents without explicitly encoding the relations between clauses. We propose a new approach that incorporates textual entailment paradigm aiming to infer the entailment relationship between the original document as the premise and the clauses or pairs described as the hypothesis. Our approach designs label-view hypothesis templates to improve ECPE by filtering out irrelevant emotion and cause clauses. Furthermore, we formulate candidate emotion-cause pairs as hypothesis statements, and define explicit multi-view symmetric templates to capture the emotion-cause relation semantics. The text entailment recognition for ECPE is finally implemented by fusing multi-view semantic information using a simplified capsule network. Our proposed model achieves state-of-the-art performance on ECPE compared to previous baselines. More importantly, this work demonstrates a novel effective way of applying the textual entailment paradigm to ECPE or clause-level causal discovery by designing multi-view hypothesis inference and information fusion.

Population-Average Brain Templates and Application to Automated Voxel-Wise Analysis Pipelines for Cynomolgus Macaque.

The growing number of neuroimaging studies of cynomolgus macaques require extending existing templates to facilitate species-specific application of voxel-wise neuroimaging methodologies. This study aimed to create population-averaged structural magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) templates for the cynomolgus macaques and apply the templates in fully automated voxel-wise analyses. We presented the development of symmetric and asymmetric MRI and DTI templates from a sample of 63 young male cynomolgus monkeys with the use of optimized template creation approaches. We also generated the associated average tissue probability maps and Diffeomorphic Anatomical Registration using Exponentiated Lie Algebra templates for use with the Statistical Parametric Mapping (SPM), as well as the average fractional anisotropy/skeleton targets for incorporation into tract-based spatial statistics (TBSS) framework. Both asymmetric and symmetric templates in a standardized coordinate space demonstrated low bias and high contrast. Fully automated processing using SPM was accomplished for all native MRI datasets and demonstrated outstanding performance regarding skull-stripping, segmentation, and normalization when using the MRI templates. Automated normalization to the DTI template was excellently achieved for all native DTI images using the TBSS pipeline. The cynomolgus MRI and DTI templates are anticipated to provide a common platform for precise single-subject data analysis and facilitate comparison of neuroimaging findings in cynomolgus monkeys across studies and sites. It is also hoped that the procedures of template creation and fully-automated voxel-wise frameworks will provide a straightforward avenue for investigating brain function, development, and neuro-psychopathological disorders in non-human primate models.

Unbiased age-specific structural brain atlases for Chinese pediatric population

In magnetic resonance (MR) imaging studies of child brain development, structural brain atlases usually serve as important references for the pediatric population, in which individual images are spatially normalized into a common or standard stereotactic space. However, the popular existing pediatric brain atlases (e.g., National Institutes of Health pediatric atlases, NIH-PD) are mostly based on MR images obtained from Caucasian populations and thus are not ideal for the characterization of the brains of Chinese children due to neuroanatomical differences related to genetic and environmental factors. Here, we use an unbiased template construction algorithm to create a set of age-specific Chinese pediatric (CHN-PD) atlases based on high-quality T1-and T2-weighted MR images from 328 cognitively normal Chinese children aged 6–12 years. The CHN-PD brain atlases include asymmetric and symmetric templates, sex-specific templates and tissue probability templates, and contain multiple age-specific templates at one-year intervals. A direct comparison of the CHN-PD and NIH-PD atlases reveals dramatic anatomical differences mainly in the bilateral frontal and parietal regions. After applying the CHN-PD and NIH-PD atlases to two independent Chinese pediatric datasets (N = 114 and N = 71), we find that the CHN-PD atlases result in significantly higher accuracy than the NIH-PD atlases in both predicting “brain age” and guiding brain tissue segmentation. These results suggest that the CHN-PD brain atlases are necessary for studies of the typical and atypical development of the Chinese pediatric population. These CHN-PD atlases have been released on the Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC) website (https://www.nitrc.org/projects/chn-pd).

Propagation phenomena for CNNs with asymmetric templates and distributed delays

The aim of this work is to study propagation phenomena for monotone and nonmonotone cellular neural networks with the asymmetric templates and distributed delays. More precisely, for the monotone case, we establish the existence of the leftward ($c_{-}^*$) and rightward ($c_{+}^*$) spreading speeds for CNNs by appealing to the theory developed in [26,27], and $c_{-}^*+c_{+}^*>0$. Especially, if cells possess the symmetric templates and the same delayed interactions, then $c_{-}^*=c_{+}^*>0$. Moreover, if the effect of the self-feedback interaction $α f'(0)$ is not less than 1, then both $c_{-}^*>0$ and $c_{+}^*>0$. For the non-monotone case, the leftward and rightward spreading speeds are investigated by using the results of the spreading speed for the monotone case and squeezing the given output function between two appropriate nondecreasing functions. It turns out that the leftward and rightward spreading speeds are linearly determinate in these two cases. We further obtain the existence and nonexistence of travelling wave solutions under the weaker conditions than those in [46, 47] and show that the spreading speed coincides with the minimal wave speed.

Growth of sexithiophene crystals on Cu(1 1 0) and Cu(1 1 0)-(2×1)O stripe phase—The influence of surface corrugation

Sexithiophene (6T) films have been grown by molecular beam deposition on two different substrates: a clean Cu(1 1 0) single crystal surface and a 1:1 Cu(1 1 0)/Cu(1 1 0)-(2×1)O stripe surface. Both substrates serve as atomically ordered two-fold symmetric templates, which induce epitaxial growth of the 6T crystallites due to surface corrugations. On both surfaces, 6T crystallizes in its low-temperature phase with (0 1 0) being the main crystallographic orientation of the 6T crystals relative to the substrate surface. Two different epitaxial growth behaviors are observed: while on the oxygen-passivated surface the long molecular axes of the 6T molecules are aligned exactly along the oxygen rows, small tilt angles relative to the densely packed copper rows are found on the clean copper surface. Classical misfit calculations seem to be insufficient to describe the observed epitaxial crystallization of sexithiophene. On the striped surface clean copper- and oxygen-passivated copper regions are offered on the same surface. The dominant crystalline fraction grows in an orientation also found on the fully reconstructed Cu(1 1 0)-(2×1)O surface where it grows exclusively in this orientation. A model that explains these experimental findings is discussed and it turns out that the sticking anisotropy is the dominating effect.

From extension of Loop's approximation scheme to interpolatory subdivisions

The minimum-supported bivariate C 2 -cubic spline on a 6-directional mesh constructed in our previous work [Chui, C.K., Jiang, Q.T., 2003. Surface subdivision schemes generated by refinable bivariate spline function vectors. Appl. Comput. Harmonic Anal. 15, 147–162] can be used to extend Loop's approximation subdivision scheme to introduce some parameter for controlling surface geometric shapes. This extension is achieved by considering matrix-valued subdivisions, resulting in subdivision templates of the same 1-ring template size as Loop's scheme, but with 2-dimensional matrix-valued weights. Another feature accomplished by considering such an extension is that the two components of the refinable vector-valued spline function can be reformulated, by taking certain linear combinations, to convert the approximation scheme to an interpolatory scheme, but at the expense of an increase in template size for the edge vertices. To maintain the 1-ring template size with guarantee of C 2 smoothness for interpolatory surface subdivisions, a non-spline solution is needed, by applying some constructive scheme such as the procedure discussed in our recent work [Chui, C.K., Jiang, Q.T., 2005b. Matrix-valued symmetric templates for interpolatory surface subdivisions I. Regular vertices. Appl. Comput. Harmonic Anal. 19, 303–339]. The main objective of this paper is to develop the corresponding matrix-valued 1-ring templates for the extraordinary vertices of arbitrary valences, for all of the three schemes mentioned above: the extended Loop approximation scheme, its conversion to an interpolatory scheme, and the non-spline 1-ring interpolatory scheme. The discrete Fourier transform (DFT) is applied to analyze the spectral properties of the corresponding subdivision matrices, assuring that the eigenvalues of the subdivision matrices satisfy certain conditions for C 1 smoothness at the extraordinary vertices for all of the three considerations in this paper.

Derivation of terrain roughness indicators via granulometries

Digital elevation models (DEMs) provide rich clues about various geophysical and geomorphologic processes. These clues include conspicuous protrusions and intrusions of foreground and background portions that testify the presence of channels and ridges in DEMs. We show an application of greyscale granulometries to characterize DEMs through shape–size complexity measures relative to symmetric rhombus, octagon and square templates. We first compute pattern spectra that measure the size distributions of protrusions and intrusions in a DEM. We then employ pattern spectra to compute probability size distribution functions of protrusions and intrusions relative to three templates. We finally compute shape–size complexity measures of DEM by employing these probability functions. To illustrate the implementation of granulometric approach to compute these measures of both background and foreground, we consider an interferometrically generated DEM of a part of Cameron Highlands of Malaysia. Hierarchical watersheds that could be decomposed from DEMs can be better classified via these measures.

Do our Noses Lead Us Away from the Scent?

In magnetic resonance (MR) imaging studies of child brain development, structural brain atlases usually serve as important references for the pediatric population, in which individual images are spatially normalized into a common or standard stereotactic space. However, the popular existing pediatric brain atlases (e.g., National Institutes of Health pediatric atlases, NIH-PD) are mostly based on MR images obtained from Caucasian populations and thus are not ideal for the characterization of the brains of Chinese children due to neuroanatomical differences related to genetic and environmental factors. Here, we use an unbiased template construction algorithm to create a set of age-specific Chinese pediatric (CHN-PD) atlases based on high-quality T1-and T2-weighted MR images from 328 cognitively normal Chinese children aged 6–12 years. The CHN-PD brain atlases include asymmetric and symmetric templates, sex-specific templates and tissue probability templates, and contain multiple age-specific templates at one-year intervals. A direct comparison of the CHN-PD and NIH-PD atlases reveals dramatic anatomical differences mainly in the bilateral frontal and parietal regions. After applying the CHN-PD and NIH-PD atlases to two independent Chinese pediatric datasets (N = 114 and N = 71), we find that the CHN-PD atlases result in significantly higher accuracy than the NIH-PD atlases in both predicting “brain age” and guiding brain tissue segmentation. These results suggest that the CHN-PD brain atlases are necessary for studies of the typical and atypical development of the Chinese pediatric population. These CHN-PD atlases have been released on the Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC) website (https://www.nitrc.org/projects/chn-pd).

CHARACTERISTIC OF MUTUALLY COUPLED TWO-LAYER CNN AND ITS STABILITY

This paper presents some interesting image processing applications with the mutually coupled two-layer Cellular Neural Networks (CNNs). We found that the two-layer CNNs are very useful compared to single layer CNNs in some applications such as center point detection, skeletonization, and so on. We also focus our discussions on both their transients and operations. In addition, the stability of the two-layer CNNs with mutually coupled symmetric templates is also discussed based on those of decoupling CNN technique.

Elementary linear filtering tasks using CNNs with minimum-size templates

In this paper we investigate the linear filtering capabilities of the standard cellular neural network in the general case of non-symmetric templates. We approached here systematically the CNNs with minimum-size templates (1x3), analyzing in detail their filtering capabilities in the one-dimensional case. Starting from a general form of the spatial transfer function, we emphasize some useful filtering functions that can be obtained. For each filter type, we derive the relations which give the template parameter values, in order to design a given CNN filter with specified characteristics-like central frequency, bandwidth, selectivity etc. Filters with symmetric templates are treated as a particular case. For each type of filtering the characteristics are shown and simulation results are presented as well. Some of these results are then extended to 2-D CNNs and several simulations of useful filtering tasks are presented on real images.

Learning of cellular neural networks

In this paper an approach to first-order cellular neural networks (CNN) learning is suggested. It is based on the ideas of perceptron learning rule. The proposed method allows to find parameters of a CNN connection template which provides the formation of patterns with preset properties. The method was applied to learn 1D and 2D CNN with symmetric templates and was verified by simulations. Experimental results are compared with the known ones theoretically obtained by Thiran et al. [P. Thiran, K. Crounse, L.O. Chua, M. Hasler, IEEE Trans. Circuits and Systems-1 42 (10) (1995)] for CNN pattern properties.

A new compact neuron-bipolar junction transistor (νBJT) cellular neural network (CNN) structure with programmable large neighborhood symmetric templates for image processing

Based on the basic device physics of the neuron-bipolar junction transistor (/spl nu/BJT), a new compact cellular neural network (CNN) structure called the /spl nu/BJT CNN is proposed and analyzed. In the /spl nu/BJT CNN, both /spl nu/BJT and lambda bipolar transistor realized by parasitic p-n-p BJTs in the CMOS process are used to implement the neuron whereas the coupling MOS resistors are used to realize the symmetric synapse weights among various neurons. Thus it has the advantages of small chip area and high integration capability. Moreover, the proposed symmetric /spl nu/BJT CNN can be easily designed to achieve large neighborhood without extra interconnection. By adding a metal-layer optical window to the /spl nu/BJT, the /spl nu/BJT can be served as the phototransistor, and the /spl nu/BJT CNN can receive optical images as initial state inputs or external inputs. The correct functions of the /spl nu/BJT CNNs in noise removal, hole filling, and erosion have been successfully verified in HSPICE simulation. An experimental chip containing a 32/spl times/32 /spl nu/BJT CNN and a 16 /spl times/16 /spl nu/BJT CNN with phototransistor design, has been designed and fabricated in 0.6-/spl mu/m single-poly triple-metal n-well CMOS technology. The fabricated chips have the cell state transition time of 0.8 /spl mu/s and the static power consumption of 60 /spl mu/W/cell. The area density can be as high as 1270 cells/mm/sup 2/. The measurement results have also confirmed the correct functions of the proposed /spl nu/BJT CNNs.