Search

Liang Chen,1 Xiaoming Liu,2 Hongjie Tang1 1School of Journalism and Communication, Tsinghua University, Beijing, People’s Republic of China; 2Department of Journalism, School of Communication, Hong Kong Baptist University, Hong Kong, People’s Republic of ChinaCorrespondence: Xiaoming Liu, Department of Journalism, School of Communication, Hong Kong Baptist University, 5 Hereford Road, Kowloon Tong, Hong Kong, People’s Republic of China, Tel +852 59826538, Email liuxiaoming@hkbu.edu.hkPurpose: Applying the parental mediation theory, this study investigates the main effects and interactive effects of three parental social media strategies, ie, active mediation, restrictive mediation, and non-intrusive inspection, on cyberbullying among teenagers. A matched child-parent survey was conducted with 642 secondary-school students aged 13– 18 and their parents in China.Results: The results showed that active mediation was negatively associated with both cyberbullying victimization and perpetration. Restrictive mediation was not significantly associated with cyberbullying. Non-intrusive inspection was positively associated with cyberbullying perpetration but not associated with cyberbullying victimization. Moreover, the two-way interaction between active and restrictive mediation as well as the three-way interaction between the three parental mediation strategies significantly affected cyberbullying perpetration and victimization. Specifically, the combination of high-level active mediation and non-intrusive inspection with low-level restrictive mediation was the most effective strategy for preventing cyberbullying.Conclusion: This finding significantly contributes to the parental mediation literature and provide theoretical guidelines for parental intervention to prevent cyberbullying among teenager.Keywords: parental mediation, social media, cyberbullying, interactive effects

Whole-body biometric recognition is an important area of research due to its vast applications in law enforcement, border security, and surveillance. This paper presents the end-to-end design, development and evaluation of FarSight, an innovative software system designed for whole-body (fusion of face, gait and body shape) biometric recognition. FarSight accepts videos from elevated platforms and drones as input and outputs a candidate list of identities from a gallery. The system is designed to address several challenges, including (i) low-quality imagery, (ii) large yaw and pitch angles, (iii) robust feature extraction to accommodate large intra-person variabilities and large inter-person similarities, and (iv) the large domain gap between training and test sets. FarSight combines the physics of imaging and deep learning models to enhance image restoration and biometric feature encoding. We test FarSight's effectiveness using the newly acquired IARPA Biometric Recognition and Identification at Altitude and Range (BRIAR) dataset. Notably, FarSight demonstrated a substantial performance increase on the BRIAR dataset, with gains of +11.82% Rank-20 identification and +11.3% TAR@1% FAR., 11 pages, 7 figures

3D sensing for monocular in-the-wild images, e.g., depth estimation and 3D object detection, has become increasingly important. However, the unknown intrinsic parameter hinders their development and deployment. Previous methods for the monocular camera calibration rely on specific 3D objects or strong geometry prior, such as using a checkerboard or imposing a Manhattan World assumption. This work solves the problem from the other perspective by exploiting the monocular 3D prior. Our method is assumption-free and calibrates the complete $4$ Degree-of-Freedom (DoF) intrinsic parameters. First, we demonstrate intrinsic is solved from two well-studied monocular priors, i.e., monocular depthmap, and surface normal map. However, this solution imposes a low-bias and low-variance requirement for depth estimation. Alternatively, we introduce a novel monocular 3D prior, the incidence field, defined as the incidence rays between points in 3D space and pixels in the 2D imaging plane. The incidence field is a pixel-wise parametrization of the intrinsic invariant to image cropping and resizing. With the estimated incidence field, a robust RANSAC algorithm recovers intrinsic. We demonstrate the effectiveness of our method by showing superior performance on synthetic and zero-shot testing datasets. Beyond calibration, we demonstrate downstream applications in image manipulation detection & restoration, uncalibrated two-view pose estimation, and 3D sensing. Codes, models, and data will be held in https://github.com/ShngJZ/WildCamera.

Geopolymer prepared from solid waste is a high value-added means. However, when used alone, the geopolymer produced by phosphogypsum has the risk of expansion cracking, while the geopolymer of recycled fine powder has high strength and good density, but its volume shrinkage and deformation are large. If the two are combined, the synergistic effect of the phosphogypsum geopolymer and recycled fine powder geopolymer can realize the complementarity of advantages and disadvantages, which provides a possibility for the preparation of stable geopolymers. In this study, the volume stability, water stability and mechanical stability of geopolymers were tested, and the stability synergy mechanism between phosphogypsum, recycled fine powder and slag was analyzed by micro experiments. The results show that the synergistic effect of phosphogypsum, recycled fine powder and slag can not only control the production of ettringite (AFt) but also control the capillary stress in the hydration product, thus improving the volume stability of the geopolymer. The synergistic effect can not only improve the pore structure of the hydration product but also reduce the negative impact of calcium sulfate dihydrate (CaSO4∙2H2O), thus improving the water stability of geopolymers. The softening coefficient of P15R45 with a 45 wt.% recycled fine powder content can reach 1.06, which is 26.2% higher than P35R25 with a 25 wt.% recycled fine powder content. The synergistic work reduces the negative impact of delayed AFt and improves the mechanical stability of the geopolymer.

Generating synthetic datasets for training face recognition models is challenging because dataset generation entails more than creating high fidelity images. It involves generating multiple images of same subjects under different factors (\textit{e.g.}, variations in pose, illumination, expression, aging and occlusion) which follows the real image conditional distribution. Previous works have studied the generation of synthetic datasets using GAN or 3D models. In this work, we approach the problem from the aspect of combining subject appearance (ID) and external factor (style) conditions. These two conditions provide a direct way to control the inter-class and intra-class variations. To this end, we propose a Dual Condition Face Generator (DCFace) based on a diffusion model. Our novel Patch-wise style extractor and Time-step dependent ID loss enables DCFace to consistently produce face images of the same subject under different styles with precise control. Face recognition models trained on synthetic images from the proposed DCFace provide higher verification accuracies compared to previous works by $6.11\%$ on average in $4$ out of $5$ test datasets, LFW, CFP-FP, CPLFW, AgeDB and CALFW. Code is available at https://github.com/mk-minchul/dcface, To appear in CVPR 2023

Numerous adversarial attack methods have been developed to generate imperceptible image perturbations that can cause erroneous predictions of state-of-the-art machine learning (ML) models, in particular, deep neural networks (DNNs). Despite intense research on adversarial attacks, little effort was made to uncover 'arcana' carried in adversarial attacks. In this work, we ask whether it is possible to infer data-agnostic victim model (VM) information (i.e., characteristics of the ML model or DNN used to generate adversarial attacks) from data-specific adversarial instances. We call this 'model parsing of adversarial attacks' - a task to uncover 'arcana' in terms of the concealed VM information in attacks. We approach model parsing via supervised learning, which correctly assigns classes of VM's model attributes (in terms of architecture type, kernel size, activation function, and weight sparsity) to an attack instance generated from this VM. We collect a dataset of adversarial attacks across 7 attack types generated from 135 victim models (configured by 5 architecture types, 3 kernel size setups, 3 activation function types, and 3 weight sparsity ratios). We show that a simple, supervised model parsing network (MPN) is able to infer VM attributes from unseen adversarial attacks if their attack settings are consistent with the training setting (i.e., in-distribution generalization assessment). We also provide extensive experiments to justify the feasibility of VM parsing from adversarial attacks, and the influence of training and evaluation factors in the parsing performance (e.g., generalization challenge raised in out-of-distribution evaluation). We further demonstrate how the proposed MPN can be used to uncover the source VM attributes from transfer attacks, and shed light on a potential connection between model parsing and attack transferability.

The multistage robust unit commitment (UC) is of paramount importance for achieving reliable operations considering the uncertainty of renewable realizations. The typical affine decision rule method and the robust feasible region method may achieve uneconomic dispatches as the dispatch decisions just rely on the current-stage information. Through approximating the future cost-to-go functions, the dual dynamic programming based methods have been shown adaptive to the multistage robust optimization problems, while suffering from high computational complexity. Thus, we propose the robust approximate dynamic programming (RADP) method to promote the computational speed and the economic performance for large-scale robust UC problems. RADP initializes the candidate points for guaranteeing the feasibility of upper bounding the value functions, solves the linear McCormick relaxation based bilinear programming to obtain the worst cases, and combines the primal and dual updates for this hybrid binary and continuous decision-making problem to achieve fast convergence. We can verify that the RADP method enjoys a finite termination guarantee for the multistage robust optimization problems with achieving suboptimal solutions. Numerical tests on 118-bus and 2383-bus transmission systems have demonstrated that RADP can approach the suboptimal economic performance at significantly improved computational efficiency., 9 pages, 3 figures, 4 tables

State-of-the-art defense mechanisms against face attacks achieve near perfect accuracies within one of three attack categories, namely adversarial, digital manipulation, or physical spoofs, however, they fail to generalize well when tested across all three categories. Poor generalization can be attributed to learning incoherent attacks jointly. To overcome this shortcoming, we propose a unified attack detection framework, namely UniFAD, that can automatically cluster 25 coherent attack types belonging to the three categories. Using a multi-task learning framework along with k-means clustering, UniFAD learns joint representations for coherent attacks, while uncorrelated attack types are learned separately. Proposed UniFAD outperforms prevailing defense methods and their fusion with an overall TDR = 94.73% @ 0.2% FDR on a large fake face dataset consisting of 341K bona fide images and 448K attack images of 25 types across all 3 categories. Proposed method can detect an attack within 3 milliseconds on a Nvidia 2080Ti. UniFAD can also identify the attack types and categories with 75.81% and 97.37% accuracies, respectively.

Supplementary Material 1

Differences in forgery attributes of images generated in CNN-synthesized and image-editing domains are large, and such differences make a unified image forgery detection and localization (IFDL) challenging. To this end, we present a hierarchical fine-grained formulation for IFDL representation learning. Specifically, we first represent forgery attributes of a manipulated image with multiple labels at different levels. Then we perform fine-grained classification at these levels using the hierarchical dependency between them. As a result, the algorithm is encouraged to learn both comprehensive features and inherent hierarchical nature of different forgery attributes, thereby improving the IFDL representation. Our proposed IFDL framework contains three components: multi-branch feature extractor, localization and classification modules. Each branch of the feature extractor learns to classify forgery attributes at one level, while localization and classification modules segment the pixel-level forgery region and detect image-level forgery, respectively. Lastly, we construct a hierarchical fine-grained dataset to facilitate our study. We demonstrate the effectiveness of our method on $7$ different benchmarks, for both tasks of IFDL and forgery attribute classification. Our source code and dataset can be found: \href{https://github.com/CHELSEA234/HiFi_IFDL}{github.com/CHELSEA234/HiFi-IFDL}., Comment: To appear at CVPR2023; 17 pages, 15 figures and 10 tables

Advancements in the generation quality of various Generative Models (GMs) has made it necessary to not only perform binary manipulation detection but also localize the modified pixels in an image. However, prior works termed as passive for manipulation localization exhibit poor generalization performance over unseen GMs and attribute modifications. To combat this issue, we propose a proactive scheme for manipulation localization, termed MaLP. We encrypt the real images by adding a learned template. If the image is manipulated by any GM, this added protection from the template not only aids binary detection but also helps in identifying the pixels modified by the GM. The template is learned by leveraging local and global-level features estimated by a two-branch architecture. We show that MaLP performs better than prior passive works. We also show the generalizability of MaLP by testing on 22 different GMs, providing a benchmark for future research on manipulation localization. Finally, we show that MaLP can be used as a discriminator for improving the generation quality of GMs. Our models/codes are available at www.github.com/vishal3477/pro_loc., Comment: Published at Conference on Computer Vision and Pattern Recognition 2023

This work studies the generalization issue of face anti-spoofing (FAS) models on domain gaps, such as image resolution, blurriness and sensor variations. Most prior works regard domain-specific signals as a negative impact, and apply metric learning or adversarial losses to remove them from feature representation. Though learning a domain-invariant feature space is viable for the training data, we show that the feature shift still exists in an unseen test domain, which backfires on the generalizability of the classifier. In this work, instead of constructing a domain-invariant feature space, we encourage domain separability while aligning the live-to-spoof transition (i.e., the trajectory from live to spoof) to be the same for all domains. We formulate this FAS strategy of separability and alignment (SA-FAS) as a problem of invariant risk minimization (IRM), and learn domain-variant feature representation but domain-invariant classifier. We demonstrate the effectiveness of SA-FAS on challenging cross-domain FAS datasets and establish state-of-the-art performance., Comment: Accepted in CVPR2023

Dense geometric matching determines the dense pixel-wise correspondence between a source and support image corresponding to the same 3D structure. Prior works employ an encoder of transformer blocks to correlate the two-frame features. However, existing monocular pretraining tasks, e.g., image classification, and masked image modeling (MIM), can not pretrain the cross-frame module, yielding less optimal performance. To resolve this, we reformulate the MIM from reconstructing a single masked image to reconstructing a pair of masked images, enabling the pretraining of transformer module. Additionally, we incorporate a decoder into pretraining for improved upsampling results. Further, to be robust to the textureless area, we propose a novel cross-frame global matching module (CFGM). Since the most textureless area is planar surfaces, we propose a homography loss to further regularize its learning. Combined together, we achieve the State-of-The-Art (SoTA) performance on geometric matching. Codes and models are available at https://github.com/ShngJZ/PMatch., Comment: CVPR2023

Feature fusion plays a crucial role in unconstrained face recognition where inputs (probes) comprise of a set of $N$ low quality images whose individual qualities vary. Advances in attention and recurrent modules have led to feature fusion that can model the relationship among the images in the input set. However, attention mechanisms cannot scale to large $N$ due to their quadratic complexity and recurrent modules suffer from input order sensitivity. We propose a two-stage feature fusion paradigm, Cluster and Aggregate, that can both scale to large $N$ and maintain the ability to perform sequential inference with order invariance. Specifically, Cluster stage is a linear assignment of $N$ inputs to $M$ global cluster centers, and Aggregation stage is a fusion over $M$ clustered features. The clustered features play an integral role when the inputs are sequential as they can serve as a summarization of past features. By leveraging the order-invariance of incremental averaging operation, we design an update rule that achieves batch-order invariance, which guarantees that the contributions of early image in the sequence do not diminish as time steps increase. Experiments on IJB-B and IJB-S benchmark datasets show the superiority of the proposed two-stage paradigm in unconstrained face recognition. Code and pretrained models are available in https://github.com/mk-minchul/caface, To appear in NeurIPS 2022

Currently, there are many publicly available Next Generation Sequencing tools developed for variant annotation and classification. However, as modern sequencing technology produces more and more sequencing data, a more efficient analysis program is desired, especially for variant analysis. In this study, we updated SNPAAMapper, a variant annotation pipeline by converting perl codes to python for generating annotation output with an improved computational efficiency and updated information for broader applicability. The new pipeline written in Python can classify variants by region (Coding Sequence, Untranslated Regions, upstream, downstream, intron), predict amino acid change type (missense, nonsense, etc.), and prioritize mutation effects (e.g., synonymous > non-synonymous) while being faster and more efficient. Our new pipeline works in five steps. First, exon annotation files are generated. Next, the exon annotation files are processed, and gene mapping and feature information files are produced. Afterward, the python scrips classify the variants based on genomic regions and predict the amino acid change category. Lastly, another python script prioritizes and ranks the mutation effects of variants to output the result file. The Python version of SNPAAMapper accomplished the overall speed by running most annotation steps in a substantially shorter time. The Python script can classify variants by region in 53 s compared to 166 s for the Perl script in a test sample run on a Latitude 7480 Desktop computer with 8GB RAM and an Intel Core i5-6300 CPU @ 2.4Ghz. Steps of predicting amino acid change type and prioritizing mutation effects of variants were executed within 1 s for both pipelines. SNPAAMapper-Python was developed and tested on the ClinVar database, a NCBI database of information on genomic variation and its relationship to human health. We believe our developed Python version of SNPAAMapper variant annotation pipeline will benefit the community by elucidating the variant consequence and speed up the discovery of causative genetic variants through whole genome/exome sequencing. Source codes, test data files, instructions, and further explanations are available on the web at https://github.com/BaiLab/SNPAAMapper-Python.

In this study, a spatial spectrum method is proposed to cope with the pilot spoofing attack (PSA) problem by exploiting the of uplink–downlink channel reciprocity in time‐division‐duplex multiple‐input multiple‐output systems. First, the spoofing attack in the uplink stage is detected by a threshold derived from the predefined false alarm based on the estimated spatial spectrum. When the PSA occurs, the transmitter (That is Alice) can detect either one or two spatial spectrum peaks. Then, the legitimate user (That is Bob) and Eve are recognized in the downlink stage via the channel reciprocity property based on the difference between the spatial spectra if PSA occurs. This way, the presence of Eve and the direction of arrival of Eve and Bob can be identified at the transmitter end. Because noise is suppressed by a spatial spectrum, the detection performance is reliable even for low signal‐noise ratios and a short training length. Consequently, Bob can use beamforming to transmit secure information during the data transmission stage. Theoretical analysis and numerical simulations are performed to evaluate the performance of the proposed scheme compared with conventional methods.

We investigated the degradation behavior of amorphous silicon nitride (Si3N4) fibers in low air pressure and presumed the evolution mechanism. The obtained Si3N4 fibers were characterized by tensile strength, X-ray diffraction, scanning electron microscopy, X-ray photoelectron spectroscopy and elemental analysis after being annealed (air pressure: 1 Pa–0.1 MPa, temperature: 1000–1600 °C, dwell time: 0–4 h). When air pressure was lower than 100 Pa or higher than 1000 Pa the strength of fibers dropped sharply. Due to the moderate partial pressure of oxygen in 100–1000 Pa, both active oxidation and passive oxidation were restrained resulting in the best mechanical property of fibers in 100–1000 Pa. Besides air pressure, annealing time also affected the thermal behavior of fibers. Firstly Si3N4 fibers were passive oxidized to form SiO2 layer on the surface, and then Si3N4 decomposed into free Si and SiO2 released gaseous SiO. Finally crystallization inside of fibers and formation of nanowires on the outer surface played the leading role in the progress of degradation.

Knowledge graph (KG), integrating complex information and containing rich semantics, is widely considered as side information to enhance the recommendation systems. However, most of the existing KG-based methods concentrate on encoding the structural information in the graph, without utilizing the collaborative signals in user-item interaction data, which are important for understanding user preferences. Therefore, the representations learned by these models are insufficient for representing semantic information of users and items in the recommendation environment. The combination of both kinds of data provides a good chance to solve this problem. To tackle this research gap, we propose a novel duet representation learning framework named \sysname to fuse local information (user-item interaction data) and global information (external knowledge graph) for the top-$N$ recommendation, which is composed of two separate sub-models. One learns the local representations by discovering the inner correlations in local information with a knowledge-aware co-attention mechanism, and another learns the global representations by encoding the knowledge associations in global information with a relation-aware attention network. The two sub-models are jointly trained as part of the semantic fusion network to compute the user preferences, which discriminates the contribution of the two sub-models under the special context. We conduct experiments on two real-world datasets, and the evaluations show that KADM significantly outperforms state-of-art methods. Further ablation studies confirm that the duet architecture performs significantly better than either sub-model on the recommendation tasks., Comment: Accepted by SIGIR 2022 as Full Paper

Background:Glioma is the most fatal neoplasm among the primary intracranial cancers. Necroptosis, a form of programmed cell death, is correlated with tumor progression and immune response. But, the role of necroptosis-related genes (NRGs) in glioma has not been well-uncovered.Methods:Single-cell and bulk RNA sequencing data, obtained from publicly accessed databases, were used to establish a necroptosis-related gene signature for predicting the prognosis of glioma patients. Multiple bioinformatics algorithms were conducted to evaluate the efficacy of the signature. The relative mRNA level of each signature gene was validated by quantitative real-time reverse transcription PCR (qRT-PCR) in glioma cell lines compared to human astrocytes.Results:In this predicted prognosis model, patients with a high risk score showed a shorter overall survival, which was verified in the testing cohorts. The signature risk score was positively related with immune cell infiltration and some immune check points, such as CD276 (B7-H3), CD152 (CTLA-4), CD223 (LAG-3), and CD274 (PD-L1). Single-cell RNA sequencing analysis confirmed that the glioma microenvironment consists of various immune cells with different markers. The eight NRGs of the signature were detected to be expressed in several immune cells. QRT-PCR results verified that all the eight signature genes were differentially expressed between human astrocytes and glioma cells.Conclusion:The eight NRGs correlate with the immune microenvironment of glioma according to our bioinformatics analysis. This necroptosis-related gene signature may evaluate the precise methodology of predicting prognosis of glioma and provide a novel thought in glioma investigation.

It is an effective method to prepare geopolymer with recycled brick and concrete fine powder (RP) and slag as main materials for the resource utilization of construction waste. However, its hydration products have large drying shrinkage and high efflorescence risk under normal curing conditions. Until now, the durability of recycled brick and concrete fine powder–slag-based geopolymer (RPSG) has not been well documented, such as drying shrinkage and efflorescence. In this study, the effects of slag content, alkali equivalent and modulus on the durability properties of RPSG were evaluated. The results show: (1) Slag can significantly reduce the drying shrinkage and efflorescence of RPSG. (2) The potential for the efflorescence of RPSG increases with increasing alkali equivalent. The drying shrinkage of RPSG increases with the increase of alkali equivalent in the case of a low alkali equivalent (6 wt.% in this paper) and decreases with the increase of alkali equivalent in the case of a high alkali equivalent. (3) The drying shrinkage of RPSG increases with increasing modulus. In contrast, the degree of efflorescence decreases with increasing modulus. In this study, RP-S45-M1.3N6 (slag content: 45 wt.%; alkali equivalent: 6 wt.%; modulus: 1.3) is the best proportional design for RPSG with excellent durability. Compared to RP-S0-M1.3N6, the drying shrinkage of RP-S45-M1.3N6 is reduced by 76.32%, the capillary porosity is reduced by 60.9%, the visual efflorescence is significantly alleviated, and the early pH value is reduced by approximately 2.0. This paper systematically analyzed the drying shrinkage pattern and the efflorescence pattern of RPSG, which has a positive significance for promoting the recycling of RP from construction waste.

Pilot spoofing attack (PSA) is a kind of active eavesdropping, which launches the identical training sequence to manipulate the channel estimation outcome during the pilot training phase. The newly proposed intelligent reflecting surface (IRS) system is vulnerable to the PSA because it relies on the pilot-assisted channel estimation methods to obtain channel statement information (CSI). To combat PSA in IRS-assisted systems, we consider two significant challenges: 1) channel distribution information (CDI) is uncertain, which will make the PSA detection methods based on the statistic feature (SF) invalid; 2) the noise is uncertain when detecting PSA in practical scenarios, which would impair the popular energy-based detectors. In this paper, we design a three-step training (TST) scheme. The effective detector could achieve reliable detection performance by examining the transmitter's received signal power levels when PSA occurs. The proposed method does not require the prior knowledge of noise variance and could obtain the CSI of both legitimate and illegitimate dyadic backscatter channels. Theoretical analysis and numerical simulations are presented further to demonstrate the efficiency of our proposed TST scheme.

ObjectiveT-cell acute lymphoblastic leukemia (T-ALL) is a rare hematological malignancy with a poor prognosis. The present study aims to identify the precise risk grouping of children with T-ALL.MethodsWe analyzed the outcomes for 105 consecutive patients treated using the Chinese Children’s Cancer Group ALL-2015 (CCCG-ALL-2015) protocol registered with the Chinese Clinical Trial Registry (ChiCTR-IPR-14005706) between 2015 and 2020 in our center. Nine out of 21 clinical and biological indicators were selected for the new scoring system based on the analysis in this study.ResultsThe 5-year overall survival (OS), event-free survival (EFS), and disease-free survival (DFS) rates for the 105 patients were 83.1 ± 4.8%, 72.4 ± 5.6%, and 78.4 ± 3.6%, respectively. Based on the new scoring system, 90 evaluable children were regrouped into low-risk (n=22), intermediate-risk (n=50), and high-risk (n=18) groups. The 5-year survival (OS, EFS, and RFS) rates for all patients in the low-risk group were 100%, significantly higher than the rates for those in the intermediate-risk group (91.2 ± 5.2%, 74.4 ± 8.6%, and 82.5 ± 6.2%, respectively) and high-risk group (59.0 ± 13.2%, 51.9 ± 12.4%, and 51.9 ± 12.4%, respectively) (all P values < 0.01).ConclusionThe CCCG-ALL-2015 program significantly improved the treatment outcomes for childhood T-ALL as compared with the CCCG-ALL-2008 protocol. Our new refined risk grouping system showed better stratification among pediatric T-ALL patients and better potential in evaluating therapeutic efficacy.

Lithium disilicate (Li2Si2O5) glass-ceramics are an ideal material for dental restoration; however, their intrinsic brittleness and low defect tolerance limit the scope of their clinical applications. In this study, Li2Si2O5 whiskers were creatively synthesized via a mild-condition hydrothermal reaction. Self-reinforced Li2Si2O5 glass-ceramics were sintered by introducing the Li2Si2O5 whiskers, and their effects on phase, microstructure, and mechanical properties were systematically studied. The crystal-growth and toughening mechanisms were also discussed. The results showed that the Li2Si2O5 whiskers played an important role in inducing crystallization, and improving the microstructure and properties of the glass-ceramics. With increasing amounts of Li2Si2O5 whiskers, the crystallinities increased slightly, and the average crystal size also increased. The microstructure was composed of crystals of bimodal size distributions, in which some large, rod-like Li2Si2O5 crystals epitaxially grew along with the whiskers, and small crystals directly crystallized from the parent glass-ceramic powders. The Li2Si2O5 glass-ceramics exhibited high flexural strength (389.5 ± 11.77 MPa, LDW3), and fracture toughness (3.46 ± 0.10 MPa·m1/2, LDW5). The improved properties were attributed mainly to crack deflection and bridge-toughening mechanisms.

Optimized nitrogen (N) management can increase N-use efficiency in intercropping systems. Legume-nonlegume intercropping systems can reduce N input by exploiting biological N fixation by legumes. Measurement of N utilization can help in dissecting the mechanisms underlying N uptake and utilization in legume-nonlegume intercropping systems. An experiment was performed with three planting patterns: monoculture maize (MM), monoculture soybean (SS), and maize-soybean relay intercropping (IMS), and three N application levels: zero N (NN), reduced N (RN), and conventional N (CN) to investigate crop N uptake and utilization characteristics. N recovery efficiency and 15N recovery rate of crops were higher under RN than under CN, and those under RN were higher under intercropping than under the corresponding monocultures. Compared with MM, IMS showed a lower soil N-dependent rate (SNDR) in 2012. However, the SNDR of MM rapidly declined from 86.8% in 2012 to 49.4% in 2014, whereas that of IMS declined slowly from 75.4% in 2012 to 69.4% in 2014. The interspecific N competition rate (NCRms) was higher under RN than under CN, and increased yearly. Soybean nodule dry weight and nitrogenase activities were respectively 34.2% and 12.5% higher under intercropping than in monoculture at the beginning seed stage. The amount (Ndfa) and ratio (%Ndfa) of soybean N2 fixation were significantly greater under IS than under SS. In conclusion, N fertilizer was more efficiently used under RN than under CN; in particular, the relay intercropping system promoted N fertilizer utilization in comparison with the corresponding monocultures. An intercropping system helps to maintain soil fertility because interspecific N competition promotes biological N fixation by soybean by reducing N input. Thus, a maize-soybean relay intercropping system with reduced N application is sustainable and environmentally friendly. Keywords: Maize-soybean relay intercropping, Reduced nitrogen, Biological nitrogen fixation, Nitrogen recovery efficiency, Soil nitrogen dependent rate

The rapid emergence of airborne platforms and imaging sensors are enabling new forms of aerial surveillance due to their unprecedented advantages in scale, mobility, deployment and covert observation capabilities. This paper provides a comprehensive overview of human-centric aerial surveillance tasks from a computer vision and pattern recognition perspective. It aims to provide readers with an in-depth systematic review and technical analysis of the current state of aerial surveillance tasks using drones, UAVs and other airborne platforms. The main object of interest is humans, where single or multiple subjects are to be detected, identified, tracked, re-identified and have their behavior analyzed. More specifically, for each of these four tasks, we first discuss unique challenges in performing these tasks in an aerial setting compared to a ground-based setting. We then review and analyze the aerial datasets publicly available for each task, and delve deep into the approaches in the aerial literature and investigate how they presently address the aerial challenges. We conclude the paper with discussion on the missing gaps and open research questions to inform future research avenues.

Additional file 1: Table S1. Primers and shRNAs used in this study.

Image manipulation detection algorithms are often trained to discriminate between images manipulated with particular Generative Models (GMs) and genuine/real images, yet generalize poorly to images manipulated with GMs unseen in the training. Conventional detection algorithms receive an input image passively. By contrast, we propose a proactive scheme to image manipulation detection. Our key enabling technique is to estimate a set of templates which when added onto the real image would lead to more accurate manipulation detection. That is, a template protected real image, and its manipulated version, is better discriminated compared to the original real image vs. its manipulated one. These templates are estimated using certain constraints based on the desired properties of templates. For image manipulation detection, our proposed approach outperforms the prior work by an average precision of 16% for CycleGAN and 32% for GauGAN. Our approach is generalizable to a variety of GMs showing an improvement over prior work by an average precision of 10% averaged across 12 GMs. Our code is available at https://www.github.com/vishal3477/proactive_IMD., Comment: Published at CVPR 2022

Recent research has shown that controllable image generation based on pre-trained GANs can benefit a wide range of computer vision tasks. However, less attention has been devoted to 3D vision tasks. In light of this, we propose a novel image-conditioned neural implicit field, which can leverage 2D supervisions from GAN-generated multi-view images and perform the single-view reconstruction of generic objects. Firstly, a novel offline StyleGAN-based generator is presented to generate plausible pseudo images with full control over the viewpoint. Then, we propose to utilize a neural implicit function, along with a differentiable renderer to learn 3D geometry from pseudo images with object masks and rough pose initializations. To further detect the unreliable supervisions, we introduce a novel uncertainty module to predict uncertainty maps, which remedy the negative effect of uncertain regions in pseudo images, leading to a better reconstruction performance. The effectiveness of our approach is demonstrated through superior single-view 3D reconstruction results of generic objects., Comment: to be published in ECCV 2022

In this work, we study multi-domain learning for face anti-spoofing(MD-FAS), where a pre-trained FAS model needs to be updated to perform equally well on both source and target domains while only using target domain data for updating. We present a new model for MD-FAS, which addresses the forgetting issue when learning new domain data, while possessing a high level of adaptability. First, we devise a simple yet effective module, called spoof region estimator(SRE), to identify spoof traces in the spoof image. Such spoof traces reflect the source pre-trained model's responses that help upgraded models combat catastrophic forgetting during updating. Unlike prior works that estimate spoof traces which generate multiple outputs or a low-resolution binary mask, SRE produces one single, detailed pixel-wise estimate in an unsupervised manner. Secondly, we propose a novel framework, named FAS-wrapper, which transfers knowledge from the pre-trained models and seamlessly integrates with different FAS models. Lastly, to help the community further advance MD-FAS, we construct a new benchmark based on SIW, SIW-Mv2 and Oulu-NPU, and introduce four distinct protocols for evaluation, where source and target domains are different in terms of spoof type, age, ethnicity, and illumination. Our proposed method achieves superior performance on the MD-FAS benchmark than previous methods. Our code and newly curated SIW-Mv2 are publicly available., Comment: To appear at ECCV 2022 as an oral presentation. The SiW-Mv2 dataset is detailed in the Appendix

Supplemental material, sj-docx-1-acr-10.1177_02841851221142552 for Comparison of MRI and CT-based radiomics for preoperative prediction of lymph node metastasis in pancreatic ductal adenocarcinoma by Piaoe Zeng, Chao Qu, Jianfang Liu, Jingjing Cui, Xiaoming Liu, Dianrong Xiu and Huishu Yuan in Acta Radiologica

Additional file 2: Table S2. Antibodies for western blot, immunoprecipitation, and immunofluorescence in this study.

It has been well recognized that neural network based image classifiers are easily fooled by images with tiny perturbations crafted by an adversary. There has been a vast volume of research to generate and defend such adversarial attacks. However, the following problem is left unexplored: How to reverse-engineer adversarial perturbations from an adversarial image? This leads to a new adversarial learning paradigm--Reverse Engineering of Deceptions (RED). If successful, RED allows us to estimate adversarial perturbations and recover the original images. However, carefully crafted, tiny adversarial perturbations are difficult to recover by optimizing a unilateral RED objective. For example, the pure image denoising method may overfit to minimizing the reconstruction error but hardly preserve the classification properties of the true adversarial perturbations. To tackle this challenge, we formalize the RED problem and identify a set of principles crucial to the RED approach design. Particularly, we find that prediction alignment and proper data augmentation (in terms of spatial transformations) are two criteria to achieve a generalizable RED approach. By integrating these RED principles with image denoising, we propose a new Class-Discriminative Denoising based RED framework, termed CDD-RED. Extensive experiments demonstrate the effectiveness of CDD-RED under different evaluation metrics (ranging from the pixel-level, prediction-level to the attribution-level alignment) and a variety of attack generation methods (e.g., FGSM, PGD, CW, AutoAttack, and adaptive attacks).

Although significant advances have been made in face recognition (FR), FR in unconstrained environments remains challenging due to the domain gap between the semi-constrained training datasets and unconstrained testing scenarios. To address this problem, we propose a controllable face synthesis model (CFSM) that can mimic the distribution of target datasets in a style latent space. CFSM learns a linear subspace with orthogonal bases in the style latent space with precise control over the diversity and degree of synthesis. Furthermore, the pre-trained synthesis model can be guided by the FR model, making the resulting images more beneficial for FR model training. Besides, target dataset distributions are characterized by the learned orthogonal bases, which can be utilized to measure the distributional similarity among face datasets. Our approach yields significant performance gains on unconstrained benchmarks, such as IJB-B, IJB-C, TinyFace and IJB-S (+5.76% Rank1)., Comment: to be published in ECCV 2022

Modern neural networks use building blocks such as convolutions that are equivariant to arbitrary 2D translations. However, these vanilla blocks are not equivariant to arbitrary 3D translations in the projective manifold. Even then, all monocular 3D detectors use vanilla blocks to obtain the 3D coordinates, a task for which the vanilla blocks are not designed for. This paper takes the first step towards convolutions equivariant to arbitrary 3D translations in the projective manifold. Since the depth is the hardest to estimate for monocular detection, this paper proposes Depth EquiVarIAnt NeTwork (DEVIANT) built with existing scale equivariant steerable blocks. As a result, DEVIANT is equivariant to the depth translations in the projective manifold whereas vanilla networks are not. The additional depth equivariance forces the DEVIANT to learn consistent depth estimates, and therefore, DEVIANT achieves state-of-the-art monocular 3D detection results on KITTI and Waymo datasets in the image-only category and performs competitively to methods using extra information. Moreover, DEVIANT works better than vanilla networks in cross-dataset evaluation. Code and models at https://github.com/abhi1kumar/DEVIANT, Comment: ECCV 2022

Additional file 1. Table S1. The inter-group consistency of the simultaneous non-contrast angiography and intraplaque hemorrhage (SNAP) imaging in identifying intraplaque hemorrhage (IPH) at different sites of intracranial and carotid arteries. Fig. S1 The agreement between simultaneous non-contrast angiography and intraplaque hemorrhage (SNAP) and high-resolution vessel wall imaging (HR-VWI) with digital subtraction angiography (DSA) in measuring the diameter of different intracranial and carotid arterial segments.

Additional file 1: Table S1. PCK of DCNN system with threshold of 2.5 mm. Table S2. PCK of DCNN system with threshold of 2 mm. Table S3. PCK of DCNN system with threshold of 1.5 mm. Table S4. PCK of DCNN system with threshold of 1 mm. Table S5. Results of parameter calculation of preoperative lower limbs in the test set. Table S6. Results of parameter calculation of TKA lower limbs in the test set. Fig S1. The measurement differences between AI predictions and ground truth in a patient with genu varum of the left knee. The figure shows the differences between AI predictions and ground truth for left lateral femoral condyle, medial femoral condyle, fossa intercondyle, lateral tibial condyle, medial tibial condyle and eminentia intercondyle in this patient. Red dot: ground truth; Blue dot: DCNN system.

This thesis contains three chapters on social and economic determinants of health. The first chapter investigates the causal effects of education on mental health. To address the endogeneity issues of education, the study uses Instrumental Variable strategy where actual years of education are instrumented by cohort-specific compulsory schooling years and its interaction with various childhood condition dummies. The estimates on the causal effects are sensitive to the choice of additional instruments. While majority of estimates suggest that there are no or only limited causal protective effects of education on mental health, the possibility of negative effects remains. The strong positive association between education and mental health could be explained by unobservable individual characteristics that affect education and mental health in the same direction., The second chapter studies whether more generous maternity leave protection in duration and payment can help decrease the incidence rates of pregnancy loss or termination (induced abortion, miscarriage and stillbirth) in low- and middle-income countries. Using a quasi-Difference-in-Differences approach, the study provides some evidence that increasing the generosity of maternity leave has the potential to reduce, in particular, the probability of induced abortion or miscarriage. The effects are stronger for Asian mothers who are working, mothers who have higher educational attainment and prestigious occupations, and mothers who were aged below 21 or above 30 at the realization of pregnancy outcomes., The last chapter explores how early life exposure to adverse shocks negatively affects the anthropometric outcomes of the second generation in Sub-Saharan African countries. The study reveals compelling evidence that mother’s experience of negative rainfall shocks, particularly excessive drought, from in-utero to early adolescence significantly increases the probability that their offsprings suffer from impaired growth during early childhood. The effects are mainly driven by households residing in the rural areas and children who are located at the lower end of anthropometric distribution. There is also evidence that mothers’ early life experience of negative rainfall shocks led to disadvantages in a variety of welfare outcomes persisting into marital households, which potentially contribute to next generation’s underperformance in early life growth measures.

State-of-the-art (SOTA) Generative Models (GMs) can synthesize photo-realistic images that are hard for humans to distinguish from genuine photos. Identifying and understanding manipulated media are crucial to mitigate the social concerns on the potential misuse of GMs. We propose to perform reverse engineering of GMs to infer model hyperparameters from the images generated by these models. We define a novel problem, "model parsing", as estimating GM network architectures and training loss functions by examining their generated images - a task seemingly impossible for human beings. To tackle this problem, we propose a framework with two components: a Fingerprint Estimation Network (FEN), which estimates a GM fingerprint from a generated image by training with four constraints to encourage the fingerprint to have desired properties, and a Parsing Network (PN), which predicts network architecture and loss functions from the estimated fingerprints. To evaluate our approach, we collect a fake image dataset with 100K images generated by 116 different GMs. Extensive experiments show encouraging results in parsing the hyperparameters of the unseen models. Finally, our fingerprint estimation can be leveraged for deepfake detection and image attribution, as we show by reporting SOTA results on both the deepfake detection (Celeb-DF) and image attribution benchmarks.

Mutations in the melanocortin 4 receptor gene (MC4R) are associated with obesity but little is known about the prevalence and impact of such mutations throughout human growth and development. We examined the MC4R coding sequence in 5,724 participants from the Avon Longitudinal Study of Parents and Children, functionally characterized all nonsynonymous MC4R variants and examined their association with anthropometric phenotypes from childhood to early adulthood. The frequency of heterozygous loss-of-function (LoF) mutations in MC4R was ~1 in 337 (0.30%), considerably higher than previous estimates. At age 18 years, mean differences in body weight, body mass index and fat mass between carriers and noncarriers of LoF mutations were 17.76 kg (95% CI 9.41, 26.10), 4.84 kg m-2 (95% CI 2.19, 7.49) and 14.78 kg (95% CI 8.56, 20.99), respectively. MC4R LoF mutations may be more common than previously reported and carriers of such variants may enter adult life with a substantial burden of excess adiposity.

Aiming at the problem that optical coherence tomography (OCT) images with low contrast and layer structure blur are difficult to be automatically layered, a new OCT detection method based on complex shearlet transform is proposed. The method utilises nearly optimal sparse approximation singular curves of multi-scale shearlet transform, and the contrast invariance of the phase congruence method. Compared with the Canny edge detector and wavelet methods, the complex shearlet-based method achieved the highest Pratt figure of merit (PFOM) value. The PFOM value of a step type edge is 0.92, and that of a pulse type edge is 0.98. Three types of OCT images were tested, including normal retinal macula area, dry age-related macular degeneration, and Stargardt disease. The experimental results show that the complex shearlet-based method can detect more layered structures of OCT images, especially the boundary between the ganglion cell layer and the inner plexiform layer that is difficult to detect, and it can detect various types of OCT images. The complex shearlet-based transform method provides an effective and general way to measure retinal OCT images.

Sustainable agricultural development is urgently required to satisfy future food demands while decreasing environmental costs. Intercropping can increase per-unit farmland productivity through a resource-efficient utilization. However, the fate of N in intercropping systems remains unclear. To study the yield advantages and the fate of N in additive maize–soybean relay intercropping (IMS) systems, we quantified crop yield, soil N transformation abilities, soil bacterial abundances, and the fate of 15N. This study was conducted using three planting patterns, namely, monoculture maize (Zea mays L.) (MM), monoculture soybean (Glycine max L. Merr.) (MS), and IMS, and two N application rates, specifically, no N and applied N (N1, 45 and 135 kg N ha−1 for MS and MM, correspondingly; and N for the IMS, which was the sum of the monocultures). Results showed that a higher per-unit farmland productivity and a lower land use intensity are attained in the intercropping system than in the corresponding monocultures. In addition, land equivalent ratio (LER) ranges from 1.85 to 2.20. Moreover, the fate of 15N showed that the N uptake and residual are the highest, whereas N loss in the IMS is the lowest among all planting patterns. Intercropping had an increased N use efficiency by increasing N utilization efficiency, rather than N uptake efficiency. The abundance of ammonia oxidizer and denitrifier indicated that IMS improves the structure of soil microorganisms. Furthermore, the transformation abilities of soil N denoted that intercropping strengthens ammonifying and nitrifying capacities to increase soil N residual while decreasing ammonia volatilization and N2O emission. Finally, the greenhouse warming potential and gas intensity of N2O were significantly lower in the IMS than in the corresponding monocultures. In summary, the IMS system provides an environmentally friendly approach to increasing farmland productivity.

The main cause for cascading failure is large-scale overload caused by localised abnormal flows. With the increase of wind power penetration, the complexity of flow distribution is strengthened. Therefore, it is necessary to study influences of site selection and permeability of wind power generation on cascading failure. An analysis method based on electric betweenness of nodes was put forward, and influences of permeability change on cascading failure on different location selection for wind power generation were researched. According to the study, it was found that when the electric betweenness was low, the overload of wind power led to declining uniformity of power flow distribution, resulting in transfer of flows to other lines. As a result, the heavy load of lines was further aggravated to promote the development of cascading failure. Simulation results of the IEEE30 system showed that with the increase of wind power penetration, the cascading failure was aggravated; and under the same permeability, the lower the electric betweenness of location selection was, the severer cascading failure would be.

In order to reveal the process of insulation destruction of circuit breaker in GIS under lightning impulse voltage, a new method of transient insulation numerical analysis is described in this paper. A combination of the electromagnetic transient analysis method and optimised charge simulation method (CSM) was put forward to calculate and analyse the transient insulation characteristics of 550 kV circuit breaker in GIS under lightning impulse voltage. The results show that, within 0–15 μ, though the voltage of circuit breaker did not reach its peak, there is severe electric field distortion caused by sharp change of voltage with a high probability of circuit breaker insulation breakdown. In addition, as for the circuit breaker insulation structure optimisation design, it is advisable to restrain the change rate of electric field strength amplitude so as to reduce the probability of insulation breakdown, rather than simply verify insulation structure by using maxi-mum electric field strength as the voltage input parameters of the electrostatic field under the power frequency in accordance with the traditional insulation analysis method, with the purpose of ensuring the circuit breaker can operate safely and reliably for a long time.

This letter proposes a novel pilot spoofing attack detection scheme by introducing an auxiliary node as a trusted user, which also participates in the uplink training process and assists to detect pilot spoofing in multiple-input single-output systems. To do so, we design an efficient three-phase uplink training method, with which a malicious user can be reliably detected. In contrast to the existing methods involving two-stage uplink–downlink training, our method relies only on the uplink training stage. Meanwhile, our new scheme detects spoofing attack without altering pilot signals, which is also beyond the competence of the existing methods. Numerical results show that our method improves the detection probability significantly.