Kaikki aineistot
Lisää
Abstract Homeobox (HOX) transcription factors, encoded by a subset of homeodomain superfamily genes, play pivotal roles in many aspects of cellular physiology, embryonic development, and tissue homeostasis. Findings over the past decade have revealed that mutations in HOX genes can lead to increased cancer predisposition, and HOX genes might mediate the effect of many other cancer susceptibility factors by recognizing or executing altered genetic information. Remarkably, several lines of evidence highlight the interplays between HOX transcription factors and cancer risk loci discovered by genome-wide association studies, thereby gaining molecular and biological insight into cancer etiology. In addition, deregulated HOX gene expression impacts various aspects of cancer progression, including tumor angiogenesis, cell autophagy, proliferation, apoptosis, tumor cell migration, and metabolism. In this review, we will discuss the fundamental roles of HOX genes in cancer susceptibility and progression, highlighting multiple molecular mechanisms of HOX involved gene misregulation, as well as their potential implications in clinical practice.
Abstract The advancement of artificial intelligence in education (AIED) has the potential to transform the educational landscape and influence the role of all involved stakeholders. In recent years, the applications of AIED have been gradually adopted to progress our understanding of students’ learning and enhance learning performance and experience. However, the adoption of AIED has led to increasing ethical risks and concerns regarding several aspects such as personal data and learner autonomy. Despite the recent announcement of guidelines for ethical and trustworthy AIED, the debate revolves around the key principles underpinning ethical AIED. This paper aims to explore whether there is a global consensus on ethical AIED by mapping and analyzing international organizations’ current policies and guidelines. In this paper, we first introduce the opportunities offered by AI in education and potential ethical issues. Then, thematic analysis was conducted to conceptualize and establish a set of ethical principles by examining and synthesizing relevant ethical policies and guidelines for AIED. We discuss each principle and associated implications for relevant educational stakeholders, including students, teachers, technology developers, policymakers, and institutional decision-makers. The proposed set of ethical principles is expected to serve as a framework to inform and guide educational stakeholders in the development and deployment of ethical and trustworthy AIED as well as catalyze future development of related impact studies in the field.
Abstract Zero-shot learning (ZSL) which aims to learn new concepts without any labeled training data is a promising solution to large-scale concept learning. Recently, many works implement zero-shot learning by transferring structural knowledge from the semantic embedding space to the image feature space. However, we observe that such direct knowledge transfer may suffer from the space shift problem in the form of the inconsistency of geometric structures in the training and testing spaces. To alleviate this problem, we propose a novel method which actualizes recurrent knowledge transfer (RecKT) between the two spaces. Specifically, we unite the two spaces into the joint embedding space in which unseen image data are missing. The proposed method provides a synthesis-refinement mechanism to learn the shared subspace structure (SSS) and synthesize missing data simultaneously in the joint embedding space. The synthesized unseen image data are utilized to construct the classifier for unseen classes. Experimental results show that our method outperforms the state-of-the-art on three popular datasets. The ablation experiment and visualization of the learning process illustrate how our method can alleviate the space shift problem. By product, our method provides a perspective to interpret the ZSL performance by implementing subspace clustering on the learned SSS.
Abstract Video object segmentation (VOS) is a critical yet challenging task in video analysis. Recently, many pixel-level matching VOS methods have achieved an outstanding performance without significant time consumption in fine-tuning. However, most of these methods pay little attention to (i) matching background pixels and (ii) optimizing discriminable embeddings between classes. To address these issues, we propose a new end-to-end trainable method, namely Triplet Matching for efficient semi-supervised Video Object Segmentation (TMVOS). In particular, we devise a new triplet matching strategy that considers both the foreground and background matching and pulls the nearest negative embedding further than the nearest positive one for every anchor. As a result, this method implicitly enlarges the distances between embeddings of different classes and thereby generates accurate matching maps. Additionally, a dual decoder is applied for optimizing the final segmentation so that the model better fits the complex background and relatively simple targets. Extensive experiments demonstrate that the proposed method achieves superior performance in terms of accuracy and running-time compared with the state-of-the-art methods. The source code is available at: https://github.com/CVisionProcessing/TMVOS.
Abstract When dealing with the optic disc and cup in the optical nerve head images, their joint segmentation confronts two critical problems. One is that the spatial layout of the vessels in the optic nerve head images is variant. The other is that the landmarks for the optic cup boundaries are spatially sparse and at small spatial scale. To solve these two problems, we propose a spatial-aware joint segmentation method by explicitly considering the spatial locations of the pixels and learning the multi-scale spatially dense features. We formulate the joint segmentation task from a probabilistic perspective, and derive a spatial-aware maximum conditional probability framework and the corresponding error function. Accordingly, we provide an end-to-end solution by designing a spatial-aware neural network. It consists of three modules: the atrous CNN module to extract the spatially dense features, the pyramid filtering module to produce the spatial-aware multi-scale features, and the spatial-aware segmentation module to predict the labels of pixels. We validate the state-of-the-art performances of our spatial-aware segmentation method on two public datasets, i.e., ORIGA and DRISHTI. Based on the segmentation masks, we quantify the cup-to-disk values and apply them to the glaucoma screening. High correlation between the cup-to-disk values and the risks of the glaucoma is validated on the dataset ORIGA.
Abstract Most existing salient object detection methods compute the saliency for pixels, patches, or superpixels by contrast. Such fine-grained contrast-based salient object detection methods are stuck with saliency attenuation of the salient object and saliency overestimation of the background when the image is complicated. To better compute the saliency for complicated images, we propose a hierarchical contour closure-based holistic salient object detection method, in which two saliency cues, i.e., closure completeness and closure reliability, are thoroughly exploited. The former pops out the holistic homogeneous regions bounded by completely closed outer contours, and the latter highlights the holistic homogeneous regions bounded by averagely highly reliable outer contours. Accordingly, we propose two computational schemes to compute the corresponding saliency maps in a hierarchical segmentation space. Finally, we propose a framework to combine the two saliency maps, obtaining the final saliency map. Experimental results on three publicly available datasets show that even each single saliency map is able to reach the state-of-the-art performance. Furthermore, our framework, which combines two saliency maps, outperforms the state of the arts. Additionally, we show that the proposed framework can be easily used to extend existing methods and further improve their performances substantially.
Abstract Learning-to-optimize is an emerging framework that leverages training data to speed up the solution of certain optimization problems. One such approach is based on the classical mirror descent algorithm, where the mirror map is modelled using input-convex neural networks. In this work, we extend this functional parameterization approach by introducing momentum into the iterations, based on the classical accelerated mirror descent. Our approach combines short-time accelerated convergence with stable long-time behavior. We empirically demonstrate additional robustness with respect to multiple parameters on denoising and deconvolution experiments.
In this work, we present a highly-sensitive refractive index sensor based on metatronic nanocircuits operating at near-infrared spectral range. The structure is designed based on simple nanorod geometry and fabricated by nanopatterning of transparent conducting oxides. The functionality of these polarization dependent metatronic nanocircuits is enhanced by applying tunable response. This feature is investigated by depositing NH2 (Amine) groups via plasma polymerization technique on top of indium-tin-oxide nanorods. The dielectric constant of Amine groups is a function of their thickness, which can be controlled by the RF power and the time duration of the applied plasma polymerization process. The resonance wavelengths of nanocircuits shift to higher wavelength, as the dielectric constant of the deposited material increases. An excellent agreement between the design and experimental results are obtained. Our metatronic based nanosensor offers a high-sensitive performance of 1587 nm/RIU with a satisfactory figure of merit for this class of sensors.
Abstract Recently, fully convolutional network (FCN) has been successfully used to locate spliced regions in synthesized images. However, all the existing FCN-based algorithms use real-valued FCN to process each channel separately. As a consequence, they fail to capture the inherent correlation between color channels and the integrity of three channels. So, in this paper, quaternion fully convolutional network (QFCN) is proposed to generalize FCN to quaternion domain by replacing real-valued conventional blocks in FCN with quaternion conventional blocks. In addition, a new color image splicing localization algorithm is proposed by combining QFCNs and superpixel (SP)-enhanced pairwise conditional random field (CRF). QFCNs consider three different versions (QFCN32, QFCN16, and QFCN8) with different up-sampling layers. The SP-enhanced pairwise CRF is used to refine the results of QFCNs. Experimental results on three publicly available datasets demonstrate that the proposed algorithm outperforms the existing algorithms including some conventional algorithms and some deep learning-based algorithms.
Recently it was demonstrated that atomic oxygen can cause the extraction of substrate atoms off metal surfaces thus generating chemically different active sites. For Ag(110) this process occurs when O2 is dosed at 175 K leading, at low coverage, to the formation of single Ag vacancies. Vacancy creation proceeds thereby via the formation of O-Ag-O complexes, which involve a local reconstruction of the surface and ignite the disruption of the Ag substrate. In this paper, we report on the details of such processes and on the isolated structures formed by the O adatoms in the limit of very low coverage. We employ scanning tunneling microscopy and density functional theory to unravel the complex structures of O/Ag(110) which are transiently present under specific reaction conditions. A variety of features such as isolated gray dots, sombreros, shallow gray and white structures oriented along [001] and [110], gray stripes, and lozenges were identified and assigned to O adatoms in different configurations. The zigzag chains interact strongly with the STM tip and are easily disrupted, giving rise to highly mobile, sombrero-shaped, isolated O adatoms also far away from the scanned area, i.e., from the current injection spot. Around 200 K, not only Ag vacancies, which are mobile with anisotropic migration, can merge together into rather complex features, but also the mobile Ag atoms are trapped by O adatoms, thus facilitating the formation of an oxygen-decorated Ag chain along [001] which ultimately induces the well-known added-row reconstruction.
Abstract The reconstruction and modification of metal surfaces upon O2 adsorption plays an important role in oxidation processes and in gauging their catalytic activity. Here, we show by employing scanning tunneling microscopy and the ab initio density functional theory that Ag atoms are extracted from pristine (110) terraces upon O2 dissociation, resulting in vacancies and in Ag-O complexes. The substrate roughening generates undercoordinated atoms and opens pathways to the Ag subsurface layer. With increasing O coverage, multiple vacancies give rise to remarkable structures. The mechanism is expected to be very general depending on the delicate interplay of energy and entropy, so that it may be active for other materials at different temperatures.
Abstract Recently it was demonstrated that atomic oxygen can cause the extraction of substrate atoms off metal surfaces thus generating chemically different active sites. For Ag(110) this process occurs when O2 is dosed at 175 K leading, at low coverage, to the formation of single Ag vacancies. Vacancy creation proceeds thereby via the formation of O-Ag-O complexes, which involve a local reconstruction of the surface and ignite the disruption of the Ag substrate. In this paper, we report on the details of such processes and on the isolated structures formed by the O adatoms in the limit of very low coverage. We employ scanning tunneling microscopy and density functional theory to unravel the complex structures of O/Ag(110) which are transiently present under specific reaction conditions. A variety of features such as isolated gray dots, sombreros, shallow gray and white structures oriented along [001] and [1 ̄10], gray stripes, and lozenges were identified and assigned to O adatoms in different configurations. The zigzag chains interact strongly with the STM tip and are easily disrupted, giving rise to highly mobile, sombrero-shaped, isolated O adatoms also far away from the scanned area, i.e., from the current injection spot. Around 200 K, not only Ag vacancies, which are mobile with anisotropic migration, can merge together into rather complex features, but also the mobile Ag atoms are trapped by O adatoms, thus facilitating the formation of an oxygen-decorated Ag chain along [001] which ultimately induces the well-known added-row reconstruction.
Abstract In situ trace element analysis of cumulus minerals may provide a clue to the parental magma from which the minerals crystallized. However, this is hampered by effects of the trapped liquid shift (TLS). In the Main Zone (MZ) of the Bushveld Complex, the Ti content in plagioclase grains shows a clear increase from core to rim, whereas most other elements [e.g. rare earth elements (REE), Zr, Hf, Pb] do not. This is different from the prominent intra-grain variation of all trace elements in silicate minerals in mafic dikes, which have a faster cooling rate. We suggest that crystal fractionation of trapped liquid occurred in the MZ of Bushveld and the TLS may have modified the original composition of the cumulus minerals for most trace elements except Ti during slow cooling. Quantitative model calculations suggest that the influence of the TLS depends on the bulk partition coefficient of the element. The effect on highly incompatible elements is clearly more prominent than that on moderately incompatible and compatible elements because of different concentration gradients between cores and rims of cumulate minerals. This is supported by the following observations in the MZ of Bushveld: (1) positive correlation between Cr, Ni and Mg# of clinopyroxene and orthopyroxene; (2) negative correlation between moderately incompatible elements (e.g. Mn and Sc in clinopyroxene and orthopyroxene; Sr, Ba and Eu in plagioclase); but (3) poor correlation between highly incompatible elements and Mg# of clinopyroxene and orthopyroxene or An# of plagioclase. Modeling suggests that the extent of the TLS for a trace element is also dependent on the initial fraction of the primary trapped liquid, with strong TLS occurring if the primary trapped liquid fraction is high. This is supported by the positive correlation between highly incompatible trace element abundances in cumulus minerals and whole-rock Zr contents. We have calculated the composition of the parental magma of the MZ of the Bushveld Complex. The compatible and moderately incompatible element contents of the calculated parental liquid are generally similar to those of the B3 marginal rocks, but different from those of the B1 and B2 marginal rocks. For the highly incompatible elements, we suggest that the use of the sample with the lowest whole-rock Zr content and the least degree of TLS is the best approach to obtain the parental magma composition. The heavy REE contents of the magma calculated from orthopyroxene are similar to those of B3 rocks and lower than those of B2 rocks. The calculated REE contents from clinopyroxene are generally significantly higher than for B2 or B3 rocks, and those from plagioclase are in the lower level of B2, but slightly higher than for B3. However, the calculated REE patterns for both clinopyroxene and plagioclase show strong negative Eu anomalies, which are at the lower level of the B2 field and within the B3 field, respectively. We suggest that Eu may be less affected by TLS than other REE owing to its higher bulk compatibility. Based on this and the fact that the calculated REE contents of the parental magma should be higher than the real magma composition owing to some degree of crystal fractionation and TLS, even for the sample with the lowest amount of trapped liquid, we propose that a B3 type liquid is the most likely parental magma to the MZ of the Bushveld Complex. In the lowermost part of the MZ, there is involvement of the Upper Critical Zone (UCZ) magma.
Abstract Understanding the full range of biodiversity patterns from local to global scales, through the study of the drivers of multiscale plant community composition and diversity, is a current goal of biogeography. A synthetic understanding of to what extent vegetation compositional patterns are produced by biotic factors, geography, or climate and how these patterns vary across scales is needed. This lack hinders prediction of the effects of climate change in global vegetation. Variation in community composition is examined in relation to climatic difference and geographic distance at hemispheric and continental scales. Vascular plants and bryophytes in thirteen mountain regions were analyzed: eight in Europe and five in North America, nine midlatitude and four oroarctic. Species composition differed between continents and between oroarctic and midlatitude regions. Patterns of paired regional similarity with distance were significant for all pairs and intercontinental pairs but not for those within Europe and North America. Climatic variables accounted for most of the variance in vegetation patterns revealed by general linear models of ordinations, but geographic variables of Moran eigenvectors and latitudinal zones were also important and significant. The effects of geography were typically twice as strong for vascular plants as for bryophytes. The importance of geography at these scales suggests that past evolutionary and ecological processes are as important as current fit to any climatic niche. Interpretation of observations of the impacts of global climate change should recognize geographic context and phylogeny, and policies to mitigate them, such as assisted migration, should be cautious.