Adjacent Features Research Articles

Constructed contrasts and manipulated experiences: the cathedral at Gerasa and its relationship with the adjacent Temple of Artemis complex

Buildings interact with, influence, and are influenced by other adjacent and nearby buildings and landscape features. Whether intentionally or not, complex spatial, architectural and experiential relationships were created when early churches were built adjacent to non-Christian religious buildings. The cathedral at Gerasa, in northern Jordan, was built adjacent to a Temple of Artemis complex and can only be satisfactorily understood by exploring both the architecture of each complex and the relationships that existed between them. This is achieved through architectural analysis in combination with a user experience-led method, focussing upon how users experienced each complex in light of the other and by considering the impact of civic and religious memory on user understanding. The comparative architectural experiences impacted user understanding of each complex and the institutions represented, while knowledge of the past was appropriated to create bridges between known non-Christian and less well-known Christian contexts and hierarchies.

Analysis and Prediction of CET4 Scores Based on Data Mining Algorithm

This paper presents the concept and algorithm of data mining and focuses on the linear regression algorithm. Based on the multiple linear regression algorithm, many factors affecting CET4 are analyzed. Ideas based on data mining, collecting history data and appropriate to transform, using statistical analysis techniques to the many factors influencing the CET‐4 test were analyzed, and we have obtained the CET‐4 test result and its influencing factors. It was found that the linear regression relationship between the degrees of fit was relatively high. We further improve the algorithm and establish a partition‐weighted K‐nearest neighbor algorithm. The K‐weighted K nearest neighbor algorithm and the partition algorithm are used in the CET‐4 test score classification prediction, and the statistical method is used to study the relevant factors that affect the CET‐4 test score, and screen classification is performed to predict when the comparison verification will pass. The weight K of the input feature and the adjacent feature are weighted, although the allocation algorithm of the adjacent classification effect has not been significantly improved, but the stability classification is better than K‐nearest neighbor algorithm, its classification efficiency is greatly improved, classification time is greatly reduced, and classification efficiency is increased by 119%. In order to detect potential risk graduating students earlier, this paper proposes an appropriate and timely early warning and preschool K‐nearest neighbor algorithm classification model. Taking test scores or make‐up exams and re‐learning as input features, the classification model can effectively predict ordinary students who have not graduated.

Joint Polarimetric-Adjacent Features Based on LCSR for PolSAR Image Classification

Image classification is a critical and important application in PolSAR image interpretation. Finding a feature extraction method, which can effectively describe the characteristics of the target, is an important basis for image classification. In addition to unique polarimetric features of PolSAR system, spatial adjacent features of image also need to be considered. So in this article, a joint polarimetric-adjacent features extraction method based on local convolution sparse representation is proposed for PolSAR image classification. Firstly, this article uses convolutional sparse representation to achieve the convolution of the image filters and the feature responses so as to achieve the effective combination of the polarimetric and adjacent information of the image. Meanwhile, construct and train the dictionary using local strategy in the original domain to avoid the high computational complexity and the confusion of different grounds caused by global dictionary. Finally, support vector machine (SVM) is used to combine the extracted features to achieve the classification. Three sets of full polarimetric data are used and the experiment results prove that the proposed method can effectively combine the polarimetric and adjacent information of data and have a good performance in PolSAR image classification.

Spatial Correlation Statistics Enable Transcriptome-Wide Characterization of RNA Structure Binding

Biologically important relationships can occur at identical transcriptomic locations or at proximal but non-overlapping sites, necessitating tools to uncover these spatial relationships. We present nearBynding, an algorithm that models spatial correlation between transcriptomic data types, such as RNA-binding protein (RBP) binding preferences, RNA structure and RNA modification. nearBynding can visualize overlapping colocalizations and adjacent features, analyze minimally processed (e.g. non-peak-called) data, and incorporate diverse, experimentally-derived or in silico predicted transcriptomic data types. By cross-correlating RBP binding and RNA structure data, we demonstrate that nearBynding recapitulates known RBP binding to structural motifs and provides novel biological insights into RBP binding preference of G-quadruplexes. nearBynding is available as an R/Bioconductor package and can run on a personal computer, making transcriptome-wide correlation of RNA features broadly accessible.

Learning to Anticipate Egocentric Actions by Imagination.

Anticipating actions before they are executed is crucial for a wide range of practical applications, including autonomous driving and robotics. In this paper, we study the egocentric action anticipation task, which predicts future action seconds before it is performed for egocentric videos. Previous approaches focus on summarizing the observed content and directly predicting future action based on past observations. We believe it would benefit the action anticipation if we could mine some cues to compensate for the missing information of the unobserved frames. We then propose to decompose the action anticipation into a series of future feature predictions. We imagine how the visual feature changes in the near future and then predicts future action labels based on these imagined representations. Differently, our ImagineRNN is optimized in a contrastive learning way instead of feature regression. We utilize a proxy task to train the ImagineRNN, i.e., selecting the correct future states from distractors. We further improve ImagineRNN by residual anticipation, i.e., changing its target to predicting the feature difference of adjacent frames instead of the frame content. This promotes the network to focus on our target, i.e., the future action, as the difference between adjacent frame features is more important for forecasting the future. Extensive experiments on two large-scale egocentric action datasets validate the effectiveness of our method. Our method significantly outperforms previous methods on both the seen test set and the unseen test set of the EPIC Kitchens Action Anticipation Challenge.

A Pixel-Level Segmentation Convolutional Neural Network Based on Deep Feature Fusion for Surface Defect Detection

Surface defect detection is very important for the quality control of product and routine maintenance of facilities, but it is still a big challenge due to the diversity and complexity of defects and environmental factors. To improve the accuracy of defect detection, we proposed a pixel-level segmentation network based on deep feature fusion for surface defect detection. The network adopts encoder-decoder structure, and it extracts low-, middle-, and high-level features via ResNet50 first. Second, by fusing adjacent feature maps at all levels and integrating the highest level feature map, multilevel feature aggregation module makes all feature maps contain context information and more details of defects. Then, multibranch decoder adopts attention modules and a multibranch structure to recover the details of defects gradually and improve the accuracy of defects segmentation. Finally, the segmentation result is produced by fusion of all branches outputs. We have evaluated the proposed network on three public data sets: MT, RSDD, and CFD. The results indicate that our proposed method outperforms the other compared methods in terms of F-measure and intersection of union (MT:73.7%, RSDD:85%, CFD:60.1%).

Blind Digital Watermarking Algorithm against Projection Transformation for Vector Geographic Data

Projection transformation is an important part of geographic analysis in geographic information systems, which are particularly common for vector geographic data. However, achieving resistance to projection transformation attacks on watermarking for vector geographic data is still a challenging task. We proposed a digital watermarking against projection transformation based on feature invariants for vector geographic data in this paper. Firstly, the features of projection transformation are analyzed, and the number of vertices, the storage order, and the storage direction of two adjacent objects are designed and used as the feature invariant to projection transformation. Then, the watermark index is calculated by the number of vertices of two adjacent objects, and the embedding rule is determined by the storage direction of two adjacent objects. Finally, the proposed scheme performs blind detection through the storage direction of adjacent features. Experimental results demonstrate that the method can effectively resist arbitrary projection transformation, which indicates the superior performance of the proposed method in comparison to the previous methods.

The Paradox of Informal Settlements Revealed in an ATLAS of Informality: Findings from Mapping Growth in the Most Common Yet Unmapped Forms of Urbanization

Informal settlements are the most common form of urbanization on the planet, accounting for one-third of the total urban form. It is expected that by the mid twenty-first century, up to three billion people will live in informal urban environments. However, we lack a consistent mapping method to pinpoint where that informality is located or how it expands. This paper presents the findings from a collection of standardized measurements of 260 informal settlements across the world. The main research goal is to identify a standard global sample of informal neighborhoods. It then focuses on mapping urban growth with remote sensing and direct mapping tools. The third stage classifies settlements based on how adjacency features such as development, topography, or bodies of water relate to their growth. The survey of growth corroborates the idea of informality as expanding geography, although at different rates than previously cited in the literature. We found peri-urban location to be a suitable estimator of informal settlement growth. This finding validates the comparison of multiple settlements to understand rates of change of urban informality worldwide. The findings here are vital to resolve important questions about the role of informal urban development in the context of accelerated global population growth.

Phonological Substitution is Motivated by Adjacent Phonological Features: Evidence from Classical Arabic

This article aims at investigating phonological substitution in classical Arabic. I hypothesise that consonantal and vocalic substitution is motivated by phonological features of adjacent consonantal or vocalic segments. Data of the study were collected from classical Arabic literary works in Aldiwan – encyclopaedia of Arabic poetry. Data were analysed in the framework of Chomsky and Halle’s SPE theory. Findings of the study have revealed that phonological features of consonantal or vocalic segments motivate other adjacent consonants to undergo a phonological substitution process in specific phonological contexts in classical Arabic. It has been revealed that the glide /w/ surfaces as /t/ when it is followed by /t/ or as /j/ when it occurs between two vowels, the first of which is high short /i/ and the second is low long /aː/, word-internally. The phoneme /t/ becomes /ṭ/ when it is preceded by /ṣ/, /ḍ/, /ṭ/ or /ð̣/ across a syllable, and it surfaces as /d/ when it is preceded by /d/, /z/ or /ð/ word-internally. It has been also found that the long vowels /aː, iː, uː/ replace glide phonemes in vocalic substitution processes when glides are adjacent to corresponding short vowels either word-internally or word-finally.

Shelf-edge delta and reef development on a mixed siliciclastic–carbonate margin, central Great Barrier Reef

ABSTRACT The Great Barrier Reef (GBR) is the world's largest extant mixed silicilastic–carbonate margin. Previous research on the Great Barrier Reef has suggested that the extensive barrier reef system may act as an impermeable barrier and limit the development of delta systems during lowstands, but sufficient geophysical data to support this hypothesis are lacking. We use dense sparker seismic and sub-bottom profiler data to better understand the structure of a large lobe-shaped feature (∼ 10 km × 10 km) on the shelf edge of the central GBR and the interactions between siliciclastic and carbonate sedimentary systems. Interpreted sparker seismic contains prograding clinoforms and suggest that the lobe-shaped feature was a river-dominated shelf-edge delta. A delta on the shelf edge implies that the presence of an exposed barrier reef was not a major impediment to deposition and that other adjacent lobe-shaped features are also deltaic deposits. The shelf-edge deltas were deposited onto a broad upper-slope terrace that allowed continued progradation and limited incision when sea level fell below the shelf edge. Delta foresets are commonly colonized by coral reefs, but the spatial and temporal relationship between reefs and some deltaic units remains unclear. The presence of multiple shelf-edge deltas that link to previously mapped Burdekin River paleo-channels indicates a complex history of sedimentation, with the Burdekin River delta migrating up to 100 km along the GBR margin during the late Quaternary. Regional bathymetric data suggest that large modern or recent shelf-edge deltas are rare on the GBR and that there was a broad range of sedimentary processes operating along the margin of the GBR during periods of low sea level.

Deeply supervised convolutional neural network for shadow detection based on a novel aerial shadow imagery dataset

Shadow detection is an essential work for remote sensing image analysis, as the presence of shadows in high resolution images not only degrades the radiometric information but also disturbs the image interpretation. In this paper, a convolutional neural network (CNN) based shadow detection framework for aerial remote sensing images is presented. We construct a publicly available Aerial Imagery dataset for Shadow Detection (AISD), which is the first aerial shadow imagery dataset, as far as we know. Based on AISD, we propose a novel Deeply Supervised convolutional neural network for Shadow Detection (DSSDNet). To solve the insufficient feature extraction problem of shadows, the DSSDNet model is designed to include two steps: (1) an encoder-decoder residual (EDR) structure is adopted to extract multi-level and discriminative shadow features; (2) a deeply supervised progressive fusion (DSPF) process is then imposed on EDR to further boost the detection performance by directly guiding the training of the network and fuse adjacent feature maps progressively. The proposed DSSDNet is compared with several state-of-the-art methods in both qualitative and quantitative analysis. Results show that the proposed DSSDNet is more accurate, and more consistent to the shape of the objects casting shadows, with the average F-score being 91.79% on the testing images.

Fingerprint liveness detection based on guided filtering and hybrid image analysis

Fingerprints are widely used for biometric recognition. However, many spoofing attacks based on an artificially made fingerprint occur. In this study, the authors propose an approach to detect fingerprint liveness which uses the guided filtering and hybrid image analysis. This study deals with the problem of ignoring the contribution that is brought by the sharp features when analysing the denoised image. The method described utilises both the enhanced sharp features and denoised features from the hybrid images to get better results. The input fingerprint is pre-processed by region of interest extraction and then is filtered by a guidance image for obtaining the denoised image. Then, histogram equalisation is introduced to eliminate the impact of illumination condition. The authors extract the co-occurrence of adjacent local binary pattern features from both the cropped images and the denoised images. Whilst concatenating both the features together to form a long feature, t-Distributed Stochastic Neighbour Embedding is applied to reduce the data dimension. The authors consider the fingerprint liveness detection as a two-class classification problem and use support vector machine with radial basis function kernel to solve this problem. The authors evaluate the experiments on three benchmark data sets. Experimental results demonstrate that the accuracy of the proposed method can outperform most of the state-of-art methods.

Multi-scale morphodynamics of an estuarine beach adjacent to a flood-tide delta: Assessing decadal scale erosion

Estuarine beaches are ubiquitous, yet understudied, coastal systems. The mixed hydrodynamic processes – such as combined tidal and wave forcing – and the influence of adjacent sedimentary features – such as flood-tide deltas (FTDs) – leads to complex morphodynamic processes. As such, the dynamics and evolution of these important coastal systems are poorly understood. This study synthesises a set of analyses conducted on an estuarine beach, in southeast Australia, over multiple temporal scales. Nearshore waves and currents were measured and used to determine the processes driving seasonal to yearly beach change between 2007 and 2010. These results were compared to decadal scale beach change determined from aerial photos between 1963 and 2006. We found the westward transport of sediment towards the inner estuary was the dominant nearshore process leading to erosion in the eastern regions of the beach (near the estuary entrance) and accretion in the west over yearly time scales. Cross-shore sediment transport occurred during winter storm swell conditions leading to erosion in more exposed sites and some limited accretion in sheltered zones, most likely due to sediment input from the flood-tide delta. However, severe storm swell events that propagated into the estuary led to a loss of sediment and erosion across the entire beach which was not recovered during the study period. It is likely that the erosion processes observed in the short- to medium-term (days to years) analyses are the conditions that led to the long-term shoreline retreat observed in the aerial photograph record. The long-term shoreline retreat since 1963 is likely a result of a negative sediment balance due to movement of the FTD and a lack of sediment input in the eastern regions of Shoal Bay. Shoreline retreat will likely continue unless increased sediment input occurs from marine sources or shoreline interventions are maintained.

AirNet: A Calibration Model for Low-Cost Air Monitoring Sensors Using Dual Sequence Encoder Networks

Air pollution monitoring has attracted much attention in recent years. However, accurate and high-resolution monitoring of atmospheric pollution remains challenging. There are two types of devices for air pollution monitoring, i.e., static stations and mobile stations. Static stations can provide accurate pollution measurements but their spatial distribution is sparse because of their high expense. In contrast, mobile stations offer an effective solution for dense placement by utilizing low-cost air monitoring sensors, whereas their measurements are less accurate. In this work, we propose a data-driven model based on deep neural networks, referred to as AirNet, for calibrating low-cost air monitoring sensors. Unlike traditional methods, which treat the calibration task as a point-to-point regression problem, we model it as a sequence-to-point mapping problem by introducing historical data sequences from both a mobile station (to be calibrated) and the referred static station. Specifically, AirNet first extracts an observation trend feature of the mobile station and a reference trend feature of the static station via dual encoder neural networks. Then, a social-based guidance mechanism is designed to select periodic and adjacent features. Finally, the features are fused and fed into a decoder to obtain a calibrated measurement. We evaluate the proposed method on two real-world datasets and compare it with six baselines. The experimental results demonstrate that our method yields the best performance.

영어 어말 설단폐쇄음의 탈락 양상: 고빈도 단어를 중심으로

Word frequency exerts a strong influence on coronal stop deletion. In general, high frequency words undergo word-final coronal stop deletion at a higher rate than low frequency words. This deletion process is undergone in consonant cluster simplification for economy of articulation. This study aims to investigate aspects of English word-final coronal stop deletion across a word boundary in high frequency words. Data were extracted from the Buckeye Corpus of Spontaneous American English Speech depending on morphological, phonological, phonetic and sociolinguistic variations. The results showed that different phonological environments favored one type of coronal stop variations such as [t], flapping, glottaling and deletion. Deletion was most prevalent for coronal stops in C/t,d/#C environment while coronal stops frequently underwent flapping in V/t,d/#V environment. In C/t,d/#C environment adjacent identical features were prohibited (Obligatory Contour Principle). More specifically, when the preceding consonants were more similar to coronal stops, the rate of coronal stop deletion was higher. The following consonants also influenced the coronal stop deletion. The rate of deletion was higher before sonorants than before obstruents. Therefore, actual education of coronal stop deletion should be done depending on various environments and factors.

Real-time Robust Object Detection Using an Adjacent Feature Fusion-based Single Shot Multibox Detector

A single shot multibox detector (SSD) is used as a baseline for many object detection networks, since it can provide sufficiently high accuracy in real time. However, it cannot deal with objects of various sizes, because features used in an SSD are not robust to multi-scale objects. To solve this problem, we present an improved feature pyramid for using multi-scale context information. The proposed feature pyramid fuses only adjacent features of the conventional SSD to achieve high accuracy without decreasing the processing speed. Our detector, with a 320×320 input, achieved 79.1% mean average precision (mAP) at 63 frames per second on a Pascal Visual Object Classes Challenge 2007 test set using a single Nvidia 1080 Ti graphics processing unit. This result shows better performance than existing SSDs.

A Semi-Fragile Reversible Watermarking for Authenticating 2D Engineering Graphics Based on Improved Region Nesting

To achieve high tampering localization precision and low distortion, a semi-fragile reversible watermarking for authenticating 2D engineering graphics is proposed based on an improved region nesting and a novel watermark generation. First, an intensive investigation is performed to the recently proposed region nesting (RN) partition. It is found that the original vertex and its mapped one cannot be guaranteed to be on the same line, which means that it still has room for improvement in term of distortion. Based on this, an improved region nesting partition (IRN) is proposed. Secondly, inspiring by the idea of soldiers parading, a novel watermark generation based on the short hash of the adjacent geometric features is developed. Then, a new coordinate system is constructed to achieve invariability of translation, scaling, rotation and entity re-arrangement. Based on the above techniques, a semi-fragile reversible watermarking for authenticating 2D engineering graphics is put forward. Experimental results and analysis show that IRN can obtain at least 15% distortion reduction compared with RN, the precision of tampering localization can be improved to a single vertex, and a certain semi-fragility in translation, scaling, rotation, and entity re-arrangement can be achieved. Furthermore, it has no file size expansion, and can obtain a good balance among imperceptibility, semi-fragility and tampering localization precision.

Pattern dependent profile distortion during plasma etching of high aspect ratio features in SiO2

As aspect ratios of features in microelectronics fabrication increase to beyond 100, transferring patterns using plasma etching into underlying materials becomes more challenging due to undesirable feature distortion such as twisting, tilting, and surface roughening. These distortions can be attributed to several causes including the randomness of reactive fluxes into features, charging, and pattern dependencies. Randomness mainly results from disparities in the fluxes of etching species into adjacent features, which can be exacerbated when reaching the etch front in high aspect ratio (HAR) features due to conduction limits. These stochastic variations in energy, angle, and sequence of the incident species into adjacent features, rather than reactor scale nonuniformities, produce many of the feature-to-feature variations in etch performance. Pattern dependent distortion results from interference between the features due to charging of the feature surfaces. The resulting electric fields act not only on the ions incident into a given feature, but also on the ions in adjacent features. With symmetric patterns, stochastic charging of the inside surfaces of features results in tilting of HAR features in random directions. However, with nominally identical neighboring features, electrical forces on ions inside the features should, in principle, cancel. Statistical variations will produce some random tilting; but on average, there is no systematic tilting. With asymmetric patterns, horizontal electric fields are generated by feature charging that point from dense (more positively charged) to sparse (less positively charged) areas of the pattern. These net electric fields deviate ions from normal incidence and produce systematic tilting.

Multi-scale Temporal Cues Learning for Video Person Re-Identification.

Temporal cues embedded in videos provide important clues for person Re-Identification (ReID). To efficiently exploit temporal cues with a compact neural network, this work proposes a novel 3D convolution layer called Multi-scale 3D (M3D) convolution layer. The M3D layer is easy to implement and could be inserted into traditional 2D convolution networks to learn multi-scale temporal cues by end-to-end training. According to its inserted location, the M3D layer has two variants, i.e., local M3D layer and global M3D layer, respectively. The local M3D layer is inserted between 2D convolution layers to learn spatial-temporal cues among adjacent 2D feature maps. The global M3D layer is computed on adjacent frame feature vectors to learn their global temporal relations. The local and global M3D layers hence learn complementary temporal cues. Their combination introduces a fraction of parameters to traditional 2D CNN, but leads to the strong multi-scale temporal feature learning capability. The learned temporal feature is fused with a spatial feature to compose the final spatial-temporal representation for video person ReID. Evaluations on four widely used video person ReID datasets, i.e., MARS, DukeMTMC-VideoReID, PRID2011, and iLIDS-VID demonstrate the substantial advantages of our method over the state-of-the art. For example, it achieves rank1 accuracy of 88.63% on MARS without re-ranking. Our method also achieves a reasonable trade-off between ReID accuracy and model size, e.g., it saves about 40% parameters of I3D CNN.

Multi-Stage Feature Constraints Learning for Age Estimation

The biometric information contained in a face image is affected by many factors such as living environment, racial differences, and genetic diversity, this complexity leads to the nonstationary of the age estimation. In order to reduce the overlap of face features between adjacent ages and improve the accuracy of age prediction, a multi-stage feature constraints learning method is proposed for face age estimation. The proposed method gradually refines the feature through three feature constraint stages. In each stage, the algorithm continuously updates the feature center of its corresponding age range, and minimizes the distance between each age feature and feature center of the corresponding age range through feature constraint. Feature constraint makes the feature distances between different individuals in the same age feature space smaller and decrease the overlap areas between adjacent age range feature spaces. Meanwhile, the feature distance of different age range feature space is enlarged. The proposed network efficiently merges the features of three stages and optimizes the mapping of feature maps to an ordered binary comparison space. Experiments show that the proposed method is able to effectively improve the discrimination between different age features, and hence to improve the accuracy of face age estimation. In addition, the proposed algorithm is simple enough to achieve fast face age estimation.