搜索结果：structure

Long documents often exhibit structure with hierarchically organized elements of different functions, such as section headers and paragraphs. Despite the omnipresence of document structure, its role in natural language processing (NLP) remains opaque. Do long-document Transformer models acquire an internal representation of document structure during pre-training? How can structural information be communicated to a model after pre-training, and how does it influence downstream performance? To answer these questions, we develop a novel suite of probing tasks to assess structure-awareness of long-document Transformers, propose general-purpose structure infusion methods, and evaluate the effects of structure infusion on QASPER and Evidence Inference, two challenging long-document NLP tasks. Results on LED and LongT5 suggest that they acquire implicit understanding of document structure during pre-training, which can be further enhanced by structure infusion, leading to improved end-task performance. To foster research on the role of document structure in NLP modeling, we make our data and code publicly available.

Large-scale vision-language pre-training has achieved significant performance in multi-modal understanding and generation tasks. However, existing methods often perform poorly on image-text matching tasks that require structured representations, i.e., representations of objects, attributes, and relations. As illustrated in Fig.~reffig:case (a), the models cannot make a distinction between ``An astronaut rides a horse" and ``A horse rides an astronaut". This is because they fail to fully leverage structured knowledge when learning representations in multi-modal scenarios. In this paper, we present an end-to-end framework Structure-CLIP, which integrates Scene Graph Knowledge (SGK) to enhance multi-modal structured representations. Firstly, we use scene graphs to guide the construction of semantic negative examples, which results in an increased emphasis on learning structured representations. Moreover, a Knowledge-Enhance Encoder (KEE) is proposed to leverage SGK as input to further enhance structured representations. To verify the effectiveness of the proposed framework, we pre-train our model with the aforementioned approaches and conduct experiments on downstream tasks. Experimen

We develop a framework for dualizing the Kolmogorov structure function $h_x(α)$, which then allows using computable complexity proxies. We establish a mathematical analogy between information-theoretic constructs and statistical mechanics, introducing a suitable partition function and free energy functional. We explicitly prove the Legendre-Fenchel duality between the structure function and free energy, showing detailed balance of the Metropolis kernel, and interpret acceptance probabilities as information-theoretic scattering amplitudes. A susceptibility-like variance of model complexity is shown to peak precisely at loss-complexity trade-offs interpreted as phase transitions. Practical experiments with linear and tree-based regression models verify these theoretical predictions, explicitly demonstrating the interplay between the model complexity, generalization, and overfitting threshold.

Spatial and temporal quantum correlations can be unified in the framework of the pseudo-density operators, and quantum causality between the involved events in an experiment is encoded in the corresponding pseudo-density operator. We study the relationship between local causal information and global causal structure. A space-time marginal problem is proposed to infer global causal structures from given marginal causal structures where causal structures are represented by the pseudo-density operators; we show that there almost always exists a solution in this case. By imposing the corresponding constraints on this solution set, we could obtain the required solutions for special classes of marginal problems, like a positive semidefinite marginal problem, separable marginal problem, etc. We introduce a space-time entropy and propose a method to determine the global causal structure based on the maximum entropy principle, which can be solved effectively by using a neural network. The notion of quantum pseudo-channel is also introduced and we demonstrate that the quantum pseudo-channel marginal problem can be solved by transforming it into a pseudo-density operator marginal problem via

Game dynamics structure (e.g., endogenous cycle motion) in human subjects game experiments can be predicted by game dynamics theory. However, whether the structure can be controlled by mechanism design to a desired goal is not known. Here, using the pole assignment approach in modern control theory, we demonstrate how to control the structure in two steps: (1) Illustrate an theoretical workflow on how to design a state-depended feedback controller for desired structure; (2) Evaluate the controller by laboratory human subject game experiments and by agent-based evolutionary dynamics simulation. To our knowledge, this is the first realisation of the control of the human social game dynamics structure in theory and experiment.

In this paper, we propose a server structure proposal and automatic performance verification technology which proposes and verifies an appropriate server structure on Infrastructure as a Service (IaaS) cloud with baremetal servers, container based virtual servers and virtual machines. Recently, cloud services have been progressed and providers provide not only virtual machines but also baremetal servers and container based virtual servers. However, users need to design an appropriate server structure for their requirements based on 3 types quantitative performances and users need much technical knowledge to optimize their system performances. Therefore, we study a technology which satisfies users' performance requirements on these 3 types IaaS cloud. Firstly, we measure performances of a baremetal server, Docker containers, KVM (Kernel based Virtual Machine) virtual machines on OpenStack with virtual server number changing. Secondly, we propose a server structure proposal technology based on the measured quantitative data. A server structure proposal technology receives an abstract template of OpenStack Heat and function/performance requirements and then creates a concrete template

The electronic structure of the $f$-based superconductor $\mathrm{CeIr_3}$ was studied by photoelectron spectroscopy. The energy distribution of the $\mathrm{Ce}~4f$ states were revealed by the $\mathrm{Ce}~3d-4f$ resonant photoelectron spectroscopy. The $\mathrm{Ce}~4f$ states were mostly distributed in the vicinity of the Fermi energy, suggesting the itinerant character of the $\mathrm{Ce}~4f$ states. The contribution of the $\mathrm{Ce}~4f$ states to the density of states (DOS) at the Fermi energy was estimated to be nearly half of that of the $\mathrm{Ir}~5d$ states, implying that the $\mathrm{Ce}~4f$ states have a considerable contribution to the DOS at the Fermi energy. The $\mathrm{Ce}~3d$ core-level and $\mathrm{Ce}~3d$ X-ray absorption spectra were analyzed based on a single-impurity Anderson model. The number of the $\mathrm{Ce}~4f$ states in the ground state was estimated to be $0.8-0.9$, which is much larger than the values obtained in the previous studies (i.e., $0-0.4$).

The idea of this project is to study the protein structure and sequence relationship using the hidden markov model and artificial neural network. In this context we have assumed two hidden markov models. In first model we have taken protein secondary structures as hidden and protein sequences as observed. In second model we have taken protein sequences as hidden and protein structures as observed. The efficiencies for both the hidden markov models have been calculated. The results show that the efficiencies of first model is greater that the second one .These efficiencies are cross validated using artificial neural network. This signifies the importance of protein secondary structures as the main hidden controlling factors due to which we observe a particular amino acid sequence. This also signifies that protein secondary structure is more conserved in comparison to amino acid sequence.

A promising route to the realization of Majorana fermions is in non-centrosymmetric superconductors, in which spin-orbit-coupling lifts the spin degeneracy of both bulk and surface bands. A detailed assessment of the electronic structure is critical to evaluate their suitability for this through establishing the topological properties of the electronic structure. This requires correct identification of the time-reversal-invariant momenta. One such material is BiPd, a recently rediscovered non-centrosymmetric superconductor which can be grown in large, high-quality single crystals and has been studied by several groups using angular resolved photoemission to establish its surface electronic structure. Many of the published electronic structure studies on this material are based on a reciprocal unit cell which is not the actual Brillouin zone of the material. We show here the consequences of this for the electronic structures and show how the inferred topological nature of the material is affected.

There have been several studies suggesting that protein structures solved by NMR spectroscopy and x-ray crystallography show significant differences. To understand the origin of these differences, we assembled a database of high-quality protein structures solved by both methods. We also find significant differences between NMR and crystal structures---in the root-mean-square deviations of the C$_α$ atomic positions, identities of core amino acids, backbone and sidechain dihedral angles, and packing fraction of core residues. In contrast to prior studies, we identify the physical basis for these differences by modelling protein cores as jammed packings of amino-acid-shaped particles. We find that we can tune the jammed packing fraction by varying the degree of thermalization used to generate the packings. For an athermal protocol, we find that the average jammed packing fraction is identical to that observed in the cores of protein structures solved by x-ray crystallography. In contrast, highly thermalized packing-generation protocols yield jammed packing fractions that are even higher than those observed in NMR structures. These results indicate that thermalized systems can pack mo

Recent surface force apparatus (SFA) and atomic force microscopy (AFM) can measure force curves between a probe and a sample surface in solvent. The force curve is thought as the solvation structure in some articles, because its shape is generally oscilltive and pitch of the oscillation is about the same as diameter of the solvent. However, it is not the solvation structure. It is only the force between the probe and the sample surface. Therefore, this brief paper presents a method for calculating the solvation structure from the force curve. The method is constructed by using integral equation theory, a statistical mechanics of liquid (Ornstein-Zernike equation coupled by hypernetted-chain closure). This method is considered to be important for elucidation of the solvation structure on a sample surface.

In a recent paper [arXiv:0804.3569], Takatoshi Nomura {\it et al.} reported a structural phase transition near 150 K in LaOFeAs and used space group "Cmma" to describe their X-ray diffraction data. However, they did not discuss how their proposed structure compares with the early neutron study by Cruz {\it et al.}[arXiv:0804.0795] where the low temperature structure of LaOFeAs was described by space group "P112/n". This caused some confusion, suggesting that there may be some disagreement on the low temperature structure of LaOFeAs as evidenced by several inquiries that we received. Here we show that the proposed structures from x-ray and neutron diffraction are basically identical. The P2/c (i.e., P112/n) cell becomes the primitive cell of the Cmma cell when the z-coordinate of the oxygen and iron are assumed to be exactly 0 and 0.5 (these numbers were reported to be -0.0057 and 0.5006 in neutron study). Our first-principles total-energy calculations suggest that the oxygen and iron atoms prefer to lie on the z=0 and 1/2 plane, respectively, supporting Cmma symmetry. However it is more convenient to describe the structural distortion in the primitive P2/c cell which makes it easie

The three-dimensional (3D) electronic structure of the hidden order compound URu$_2$Si$_2$ in a paramagnetic phase was revealed using a 3D angle-resolved photoelectron spectroscopy where the electronic structure of the entire Brillouin zone is obtained by scanning both incident photon energy and detection angles of photoelectrons. The quasi-particle bands with enhanced contribution from the $\mathrm{U}~5f$ state were observed near $E_\mathrm{F}$, formed by the hybridization with the $\mathrm{Ru}~4d$ states. The energy dispersion of the quasi-particle band is significantly depend on $k_z$, indicating that they inherently have a 3D nature. The band-structure calculation qualitatively explain the characteristic features of the band structure and Fermi surface although the electron correlation effect strongly renormalizes the quasi-particle bands. The 3D and strongly-correlated nature of the quasi-particle bands in URu$_2$Si$_2$ is an essential ingredient for modeling its hidden-order transition.

Sunspots contain multiple small-scale structures in the umbra and in the penumbra. Despite extensive research on this subject in pre-Hinode era multiple questions concerning fine-scale structures of sunspots, their formation, evolution and decay remained open. Several of those questions were proposed to be pursued by Hinode (SOT). Here we review some of the achievements on understanding sunspot structure by Hinode in its first 10 years of successful operation. After giving a brief summary and updates on the most recent understanding of sunspot structures, and describing contributions of Hinode to that, we also discuss future directions. This is a section (\#7.1) of a long review article on the achievements of Hinode in the first 10 years.

The Swapping Autoencoder achieved state-of-the-art performance in deep image manipulation and image-to-image translation. We improve this work by introducing a simple yet effective auxiliary module based on gradient reversal layers. The auxiliary module's loss forces the generator to learn to reconstruct an image with an all-zero texture code, encouraging better disentanglement between the structure and texture information. The proposed attribute-based transfer method enables refined control in style transfer while preserving structural information without using a semantic mask. To manipulate an image, we encode both the geometry of the objects and the general style of the input images into two latent codes with an additional constraint that enforces structure consistency. Moreover, due to the auxiliary loss, training time is significantly reduced. The superiority of the proposed model is demonstrated in complex domains such as satellite images where state-of-the-art are known to fail. Lastly, we show that our model improves the quality metrics for a wide range of datasets while achieving comparable results with multi-modal image generation techniques.

An effective structure helps an article to convey its core message. The optimal structure depends on the information to be conveyed and the expectations of the audience. In the current increasingly interdisciplinary era, structural norms can be confusing to the authors, reviewers and audiences of scientific articles. Despite this, no prior study has attempted to assess variations in the structure of academic papers across all disciplines. This article reports on the headings commonly used by over 1 million research articles from the PubMed Central Open Access collection, spanning 22 broad categories covering all academia and 172 out of 176 narrow categories. The results suggest that no headings are close to ubiquitous in any broad field and that there are substantial differences in the extent to which most headings are used. In the humanities, headings may be avoided altogether. Researchers should therefore be aware of unfamiliar structures that are nevertheless legitimate when reading, writing and reviewing articles.

The usual quantization of a classical space-time field does not touch the non-geometrical character of quantum mechanics. We believe that the deep problems of unification of general relativity and quantum mechanics are rooted in this poor understanding of the geometrical character of quantum mechanics. In Einstein's theory gravitation is expressed by geometry of space-time, and the solutions of the field equation are invariant w.r.t. a certain equivalence class of reference frames. This class can be characterized by the differential structure of space-time. We will show that matter is the transition between reference frames that belong to different differential structures, that the set of transitions of the differential structure is given by a Temperley-Lieb algebra which is extensible to a $C^{*}$-algebra comprising the field operator algebra of quantum mechanics and that the state space of quantum mechanics is the linear space of the differential structures. Furthermore we are able to explain the appearance of the complex numbers in quantum theory. The strong relation to Loop Quantum Gravity is discussed in conclusion.

Recent experimental and theoretical ideas are laying the ground for a new era in the knowledge of the parton structure of nuclei. We report on two promising directions beyond inclusive deep inelastic scattering experiments, aimed at, among other goals, unveiling the three dimensional structure of the bound nucleon. The 3D structure in coordinate space can be accessed through deep exclusive processes, whose non-perturbative content is parametrized in terms of generalized parton distributions. In this way the distribution of partons in the transverse plane will be obtained, providing a pictorial view of the realization of the European Muon Collaboration effect. In particular, we show how, through the generalized parton distribution framework, non nucleonic degrees of freedom in nuclei can be unveiled. Analogously, the momentum space 3D structure can be accessed by studying transverse momentum dependent parton distributions in semi-inclusive deep inelastic scattering processes. The status of measurements is also summarized, in particular novel coincidence measurements at high luminosity facilities, such as Jefferson Laboratory. Finally the prospects for the next years at future facili

We determine the second fundamental form of a variation of Hodge Structure of a smooth projective hypersurface using the classical identification of the Hodge structure and the action of the infinitesimal variation of Hodge structure with pieces of the Jacobian Ideal of the hypersurface. This simple algebraic description also implies that, in this case, the variation of Hodge structure satisfies a new set of second order partial differential equations.Comments are welcome!