scholarly journals Human Texture Vision as Multi-Order Spectral Analysis

2021 ◽  
Vol 15 ◽  
Author(s):  
Kosuke Okada ◽  
Isamu Motoyoshi

Texture information plays a critical role in the rapid perception of scenes, objects, and materials. Here, we propose a novel model in which visual texture perception is essentially determined by the 1st-order (2D-luminance) and 2nd-order (4D-energy) spectra. This model is an extension of the dimensionality of the Filter-Rectify-Filter (FRF) model, and it also corresponds to the frequency representation of the Portilla-Simoncelli (PS) statistics. We show that preserving two spectra and randomizing phases of a natural texture image result in a perceptually similar texture, strongly supporting the model. Based on only two single spectral spaces, this model provides a simpler framework to describe and predict texture representations in the primate visual system. The idea of multi-order spectral analysis is consistent with the hierarchical processing principle of the visual cortex, which is approximated by a multi-layer convolutional network.

2020 ◽  
Vol 34 (10) ◽  
pp. 13953-13954
Author(s):  
Xu Wang ◽  
Shuai Zhao ◽  
Bo Cheng ◽  
Jiale Han ◽  
Yingting Li ◽  
...  

Multi-hop question answering models based on knowledge graph have been extensively studied. Most existing models predict a single answer with the highest probability by ranking candidate answers. However, they are stuck in predicting all the right answers caused by the ranking method. In this paper, we propose a novel model that converts the ranking of candidate answers into individual predictions for each candidate, named heterogeneous knowledge graph based multi-hop and multi-answer model (HGMAN). HGMAN is capable of capturing more informative representations for relations assisted by our heterogeneous graph, which consists of multiple entity nodes and relation nodes. We rely on graph convolutional network for multi-hop reasoning and then binary classification for each node to get multiple answers. Experimental results on MetaQA dataset show the performance of our proposed model over all baselines.


2018 ◽  
Vol 838 ◽  
pp. 658-689 ◽  
Author(s):  
Zixuan Yang ◽  
Bing-Chen Wang

In this paper, we study the scales and dynamics of Taylor–Görtler (TG) vortices in streamwise-rotating turbulent channel flows at moderate and high rotation numbers ($Ro_{\unicode[STIX]{x1D70F}}=7.5$, 15, 30, 75 and 150) with a fixed Reynolds number. In order to precisely capture TG vortices in the streamwise and spanwise directions, direct numerical simulations have been performed on 15 test cases of different domain sizes and rotation numbers. A two-layer pattern of TG vortices is identified, and the characteristic length scales of TG vortices are quantified using the premultiplied energy spectra. It is observed that as the rotation number increases, the spanwise scale of TG vortices remains stable but the streamwise scale increases rapidly. Three criteria have been used for judging a domain-size-independent solution in both physical and spectral spaces. The weakest criterion ensures accurate predictions of the first- and second-order statistical moments of the velocity, which requires a minimum streamwise domain size of $L_{1}=64\unicode[STIX]{x03C0}h$, where $h$ is one-half the channel height. However, the streamwise domain size needs to be stretched drastically to $L_{1}=512\unicode[STIX]{x03C0}h$ if the most stringent criterion is considered, which demands that all energetic eddies be fully captured based on a predefined threshold value (i.e. $12.5\,\%$ of the peak value) of the premultiplied two-dimensional energy spectrum. The effects of streamwise system rotation on the scales and dynamics of TG vortices are investigated by comparing the statistical results of rotating and non-rotating channel flows, and through the analysis of two-point correlations, premultiplied energy spectra, and budget balance of turbulent stresses.


Blood ◽  
2006 ◽  
Vol 108 (11) ◽  
pp. 5188-5188
Author(s):  
Fan Yang ◽  
Yan Yan ◽  
Zeyu Xiong ◽  
Irene H. Chen ◽  
Hong Wang ◽  
...  

Abstract Alternative splicing is a process that removes introns and alters exons to generate multiple isoforms from a single pre-mRNA transcript. Alternative splicing is the major mechanism by which a small number of human genes (6 × 104) can encode the larger complexity of the human proteome (1 × 106 proteins). Previously we demonstrated that alternative splicing of apoptosis-regulatory protein transcripts regulates immune responses by modulating lymphocyte survival (Immunity, 1997; Mol. Immunol. 2002; J Exp Med, 2002; Oncogene 2005; Biochem J, 2005). To examine the hypothesis that alternative splicing plays a role in selection of nonmutated self-protein isoforms for tumor antigens and autoantigens, recently, we showed that alternative splicing is a major mechanism in regulation of the immunogenicity of tumor antigen CML66 (J. Immunol. 2004). In addition, we found that alternative splicing occurs in 100% of the autoantigen transcripts. This is significantly higher than the approximately 42% rate of alternative splicing observed in the 10,000 randomly selected human gene transcripts (p<0.001) [J. Allergy Clin. Immunol., 2004 (cover article)]. Here, we report that essential alternative splicing factor ASF/SF2 expression in samples from patients with chronic inflammation is lower than that of the healthy controls (p<0.05). In addition, TNF-a significantly downregulates ASF/SF2 expression (7 folds) in cultured cells in comparison to the expression variations of b-actin control. These findings demonstrate that ASF/SF2, presumably affecting splicing of self-antigen transcripts, is downregulated in autoimmune inflammatory disease potentially via a TNF-a-mediated pathway. Collectively, we propose for the first time a novel model of “stimulation-responsive splicing”, which emphasizes that stimulation-responsive splicing plays a critical role in selection of nonmutated self-protein isoforms to become tumor antigens and autoantigens (Clin. Immunol. Invited Review, in press, 2006). The new model for the definition of immunogenic isoforms of tumor antigens and autoantigens is significant in facilitating the development of: immunogenic antigen isoform microarrays for disease diagnosis and prognosis; autoantigen-tolerizing therapy and splicing-redirection therapy for autoimmune diseases; and immunogenic antigen isoforms-based immunotherapy for tumors.


Geophysics ◽  
2014 ◽  
Vol 79 (2) ◽  
pp. D91-D98 ◽  
Author(s):  
Wu Wensheng ◽  
Niu Wei ◽  
Luo Li

Because of the effects of background and measurement environment and multiplet effects of different elements, high-precision analysis of mixed capture [Formula: see text]-ray energy spectra of complicated formations remains challenging for geochemical elemental logging. The direct demodulation (DD) method makes full use of the measured data information, enabling physical constraints to be rationally applied to the spectral analysis process, and can yield high-precision elemental content from poor-statistics, low signal-to-noise ratio, and disturbed data. We construct mixed formations of different sandstones and limestones, mixed formations of sandstone and anhydrite, and more complicated mixed formations of multiple lithologies and employ Monte Carlo numerical simulations to obtain the neutron-capture [Formula: see text]-ray energy spectra of these mixed formations. We then employ the DD method and the weighted-least-squares (WLS) method to analyze quantitatively such mixed spectra, respectively, and compare the results with the actual contents of formation elements. The results indicate that the DD method offers higher precision spectral analysis compared with the results of the WLS method. The results for the capture [Formula: see text]-ray energy spectra of the formation for two actual wells also indicate that the DD method can be useful for spectral analysis in actual application.


Water ◽  
2019 ◽  
Vol 11 (8) ◽  
pp. 1538 ◽  
Author(s):  
Maoxing Wei ◽  
Nian-Sheng Cheng ◽  
Yee-Meng Chiew ◽  
Fengguang Yang

This paper presents an experimental study on the characteristics of the propeller-induced flow field and its associated scour hole around a closed type quay (with a vertical quay wall). An “oblique particle image velocimetry” (OPIV) technique, which allows a concurrent measurement of the velocity field and scour profile, was employed in measuring the streamwise flow field (jet central plane) and the longitudinal centerline scour profile. The asymptotic scour profiles obtained in this study were compared with that induced by an unconfined propeller jet in the absence of any berthing structure, which demonstrates the critical role of the presence of the quay wall as an obstacle in shaping the scour profile under the condition of different wall clearances (i.e., longitudinal distance between propeller and wall). Moreover, by comparing the vortical structure within the asymptotic scour hole around the vertical quay wall with its counterpart in the case of an open quay (with a slope quay wall), the paper examines the effect of quay types on the formation of the vortex system and how it determines the geometrical characteristic of the final scour profile. Furthermore, the temporal development of the mean vorticity field and the vortex system are discussed in terms of their implications on the evolution of the scour hole. In particular, comparison of the circulation development of the observed vortices allows a better understanding of the vortex scouring mechanism. Energy spectra analysis reveals that at the vortex centers, their energy spectra distributions consistently follow the −5/3 law throughout the entire scouring process. As the scour process evolves, the turbulent energy associated with the near-bed vortex, which is responsible for scouring, is gradually reduced, especially for the small-scale eddies, indicating a contribution of the dissipated turbulent energy in excavating the scour hole. Finally, a comparison of the near-bed flow characteristics of the average kinetic energy (AKE), turbulent kinetic energy (TKE), and Reynolds shear stress (RSS) are also discussed in terms of their implications for the scour hole development.


2021 ◽  
Vol 65 (2-4) ◽  
pp. 196-200
Author(s):  
Francesco S. Ciani ◽  
Paolo Bonfiglio ◽  
Stefano Piva

Plumes fires are characterized by a turbulent nature with a large number of different scales. LES is used to solve the largest structures and to model the smallest ones. Grid size and time steps become decisive to place a limit between solved and modelled turbulence. A spectral analysis, both in frequency and wavenumber domain of the specific turbulent kinetic energy is an instrument to check for the information investigated. Unfortunately, the spectra in the wavenumber domain can be difficult to achieve adequately, because the specific turbulent kinetic energy values should be available in many points. This issue can be overcome by identifying a correlation law between frequencies and wavenumbers. An approach to identify this correlation law can be to adopt the IWC method. Here, for a test case of a turbulent reacting plume of burning propane, specific turbulent kinetic energy is analysed both in frequency and wavenumber and a correlation law between them is identified by using the IWC method. A study has been performed to evaluate the grid dependency of the specific turbulent kinetic energy spectra, by assessing the extension of the Kolmogorov power law region. The correlation results are discussed and compared with the Taylor’s hypothesis.


2021 ◽  
Vol 2 (1) ◽  
pp. 101-105
Author(s):  
Runyu Hong ◽  
Wenke Liu ◽  
David Fenyö

Studies have shown that STK11 mutation plays a critical role in affecting the lung adenocarcinoma (LUAD) tumor immune environment. By training an Inception-Resnet-v2 deep convolutional neural network model, we were able to classify STK11-mutated and wild-type LUAD tumor histopathology images with a promising accuracy (per slide AUROC = 0.795). Dimensional reduction of the activation maps before the output layer of the test set images revealed that fewer immune cells were accumulated around cancer cells in STK11-mutation cases. Our study demonstrated that deep convolutional network model can automatically identify STK11 mutations based on histopathology slides and confirmed that the immune cell density was the main feature used by the model to distinguish STK11-mutated cases.


Author(s):  
Frank Edughom Ekpar

This paper establishes the foundational principles and practice for a unified theory of arbitrary information management by disclosing systems, devices and methods for the management of substrates or biological substrates. In this context, a substrate is any aspect of any entity that is capable of responding to or emitting stimuli irrespective of whether the stimuli actually emanate from any aspect of the entity or not. Management of substrates could be achieved through the management of stimuli that modulate or moderate or influence any aspect of the substrate as well as through the management of any stimuli emanating from the substrate. The results enable a wide range of novel applications in a variety of fields with far-reaching implications. For example, the functional organization of many regions of the brain including the superior temporal cortex which is believed to play a critical role in the hierarchical processing of human visual and auditory stimuli is poorly understood. It is not known precisely which layer within which region of the brain is responsible for which aspect of visual or auditory processing. Simultaneous non-invasive acquisition of bio-signals representing contributions from multiple layers of neuronal populations within the brain could provide new insights leading to the resolution of many of these outstanding issues and provide a deeper understanding of the underlying physiological processes.


Sign in / Sign up

Export Citation Format

Share Document