Search

CAPTCHA (Completely Automated Public Turing Test to Tell Computers and Humans Apart) has long been employed to combat automated bots. It accomplishes this by utilizing distortion techniques and cognitive characteristics. When it comes to countering security attacks, cognitive CAPTCHA methods have proven to be more effective than other approaches. The advancement of eye-tracking technology has greatly improved human-computer interaction (HCI), enabling users to engage with computers without physical contact. This technology is widely used for studying attention, cognitive processes, and performance. In this specific research, we conducted eye-tracking experiments on participants to investigate how their visual behavior changes as the complexity of cognitive CAPTCHAs varies. By analyzing the distribution of eye gaze on each level of CAPTCHA, we can assess users’ visual behavior based on eye movement performance and process metrics. The data collected is then employed in Machine Learning (ML) algorithms to categorize and examine the relative importance of these factors in predicting performance. This study highlights the potential to enhance any cognitive CAPTCHA model by gaining insights into the underlying cognitive processes.

To counteract rising bots, many CAPTCHAs (Completely Automated Public Turing tests to tell Computers and Humans Apart) have been developed throughout the years. Automated attacks, however, employing powerful deep learning techniques, have had high success rates over common CAPTCHAs, including image-based and text-based CAPTCHAs. Optimistically, introducing imperceptible noise, Adversarial Examples have lately been shown to particularly impact DNN (Deep Neural Network) networks. The authors improved the CAPTCHA security architecture by increasing the resilience of Adversarial Examples when combined with Neural Style Transfer. The findings demonstrated that the proposed approach considerably improves the security of ordinary CAPTCHAs.

Abstract Yield estimation (YE) of the crop is one of the main tasks in fruit management and marketing. Based on the results of YE, the farmers can make a better decision on the harvesting period, prevention strategies for crop disease, subsequent follow-up for cultivation practice, etc. In the current scenario, crop YE is performed manually, which has many limitations such as the requirement of experts for the bigger fields, subjective decisions and a more time-consuming process. To overcome these issues, an intelligent YE system was proposed which detects, localizes and counts the number of tomatoes in the field using SegNet with VGG19 (a deep learning-based semantic segmentation architecture). The dataset of 672 images was given as an input to the SegNet with VGG19 architecture for training. It extracts features corresponding to the tomato in each layer and detection was performed based on the feature score. The results were compared against the other semantic segmentation architectures such as U-Net and SegNet with VGG16. The proposed method performed better and unveiled reasonable results. For testing the trained model, a case study was conducted in the real tomato field at Manapparai village, Trichy, India. The proposed method portrayed the test precision, recall and F1-score values of 89.7%, 72.55% and 80.22%, respectively along with reasonable localization capability for tomatoes.

The need for reliable wireless communication in remote areas has led to the adoption of unmanned aerial vehicles (UAVs) as flying base stations (FlyBSs). FlyBSs hover over a designated area to ensure continuous communication coverage for mobile users on the ground. Moreover, rate-splitting multiple access (RSMA) has emerged as a promising interference management scheme in multi-user communication systems. In this paper, we investigate an RSMA-enhanced FlyBS downlink communication system and formulate an optimization problem to maximize the sum-rate of users, taking into account the three-dimensional FlyBS trajectory and RSMA parameters. To address this continuous non-convex optimization problem, we propose a TD3-RFBS optimization framework based on the twin-delayed deep deterministic policy gradient (TD3). This framework overcomes the limitations associated with the overestimation issue encountered in the deep deterministic policy gradient (DDPG), a well-known deep reinforcement learning method. Our simulation results demonstrate that TD3-RFBS outperforms existing solutions for FlyBS downlink communication systems, indicating its potential as a solution for future wireless networks.

Anomaly detection in video surveillance is a highly developed subject that is attracting increased attention from the research community. There is great demand for intelligent systems with the capacity to automatically detect anomalous events in streaming videos. Due to this, a wide variety of approaches have been proposed to build an effective model that would ensure public security. There has been a variety of surveys of anomaly detection, such as of network anomaly detection, financial fraud detection, human behavioral analysis, and many more. Deep learning has been successfully applied to many aspects of computer vision. In particular, the strong growth of generative models means that these are the main techniques used in the proposed methods. This paper aims to provide a comprehensive review of the deep learning-based techniques used in the field of video anomaly detection. Specifically, deep learning-based approaches have been categorized into different methods by their objectives and learning metrics. Additionally, preprocessing and feature engineering techniques are discussed thoroughly for the vision-based domain. This paper also describes the benchmark databases used in training and detecting abnormal human behavior. Finally, the common challenges in video surveillance are discussed, to offer some possible solutions and directions for future research.

Many organelles inside and outside a living cell depend on the perfect behavior of Golgi apparatus for smooth and normal functioning. Its poor performance may lead to many inheritable diseases like diabetes and cancer. Therefore, it is highly crucial to detect any strange behavior of Golgi apparatus in advance. Accurate discrimination of cis-Golgi from trans-Golgi proteins surely helps researchers identify the role of Golgi proteins in various diseases and assist pharmacists in drug development. In this work, various hybrid models of Bi-Profile Bayes, Bigram PSSM, Di-Peptide Composition, and Split Amphiphilic Pseudo Amino Acid Composition with SMOTE oversampling technique have been employed to discriminate Golgi protein types. Multiple linear Support Vector Machines have been used to exploit the discrimination power of these models. The proposed prediction system: Golgi-predictor has shown significant performance and achieved promising results compared to other existing state-of-the-art techniques. Through the 10-fold cross-validation, the proposed system achieved an accuracy value of 97.6%, sensitivity value of 98.8%, specificity value of 96.5%, G-mean value of 97.6%, MCC value of 0.95, and F-score value of 0.97. Similarly, through the jackknife cross-validation, the achieved values for accuracy, sensitivity, specificity, G-mean, MCC, and F-score are respectively, 96.5%, 97.8%, 95.2%, 96.4%, 0.93, and 0.96. Moreover, through the independent dataset testing, Golgi-predictor demonstrated significant enhancement in performance over other techniques. The proposed methodology aims at supporting drug designers in pharmaceutical industry and assisting researchers from the fields of bioinformatics and computational biology towards better innovation in predicting the behavior of Golgi proteins.

In smart agriculture, rice variety inspection systems based on computer vision need to be used for recognizing rice seeds instead of using technical experts. In this paper, we have investigated three types of local descriptors, such as Local Binary Pattern (LBP), Histogram of Oriented Gradients (HOG) and GIST to characterize rice seed images. However, this approach raises the curse of dimensionality phenomenon and needs to select the relevant features for a compact and better representation model. A new ensemble feature selection is proposed to represent all useful information collected from different single feature selection methods. The experimental results have shown the efficiency of our proposed method in terms of accuracy.

The annotated dataset is an essential requirement to develop an artificial intelligence (AI) system effectively and expect the generalization of the predictive models and to avoid overfitting. Lack of the training data is a big barrier so that AI systems can broaden in several domains which have no or missing training data. Building these datasets is a tedious and expensive task and depends on the domains and languages. This is especially a big challenge for low-resource languages. In this paper, we experiment and evaluate many various approaches on sentiment analysis problems so that they can still obtain high performances under limited training data. This paper uses the preprocessing techniques to clean and normalize the data and generate the new samples from the limited training dataset based on many text augmentation techniques such as lexicon substitution, sentence shuffling, back translation, syntax-tree transformation, and embedding mixup. Several experiments have been performed for both well-known machine learning-based classifiers and deep learning models. We compare, analyze, and evaluate the results to indicate the advantage and disadvantage points of the techniques for each approach. The experimental results show that the data augmentation techniques enhance the accuracy of the predictive models; this promises that smart systems can be applied widely in several domains under limited training data.

CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart), or HIP (Human Interactive Proof), has long been utilized to avoid bots manipulating web services. Over the years, various CAPTCHAs have been presented, primarily to enhance security and usability against new bots and cybercriminals carrying out destructive actions. Nevertheless, automated attacks supported by ML (Machine Learning), CNN (Convolutional Neural Network), and DNN (Deep Neural Network) have successfully broken all common conventional schemes, including text- and image-based CAPTCHAs. CNN/DNN have recently been shown to be extremely vulnerable to adversarial examples, which can consistently deceive neural networks by introducing noise that humans are incapable of detecting. In this study, the authors improve the security for CAPTCHA design by combining text-based, image-based, and cognitive CAPTCHA characteristics and applying adversarial examples and neural style transfer. Comprehend usability and security assessments are performed to evaluate the efficacy of the improvement in CAPTCHA. The results show that the proposed CAPTCHA outperforms standard CAPTCHAs in terms of security while remaining usable. Our work makes two major contributions: first, we show that the combination of deep learning and cognition can significantly improve the security of image-based and text-based CAPTCHAs; and second, we suggest a promising direction for designing CAPTCHAs with the concept of the proposed CAPTCHA.

We examine a general wireless sensor network (WSN) model which incorporates a large number of sensors distributed over a large and complex geographical area. The study proposes solutions for a flexible deployment, low cost and high reliability in a wireless sensor network. To achieve these aims, we propose the application of an unmanned aerial vehicle (UAV) as a flying relay to receive and forward signals that employ nonorthogonal multiple access (NOMA) for a high spectral sharing efficiency. To obtain an optimal number of subclusters and optimal UAV positioning, we apply a sensor clustering method based on K-means unsupervised machine learning in combination with the gap statistic method. The study proposes an algorithm to optimize the trajectory of the UAV, i.e., the centroid-to-next-nearest-centroid (CNNC) path. Because a subcluster containing multiple sensors produces cochannel interference which affects the signal decoding performance at the UAV, we propose a diagonal matrix as a phase-shift framework at the UAV to separate and decode the messages received from the sensors. The study examines the outage probability performance of an individual WSN and provides results based on Monte Carlo simulations and analyses. The investigated results verified the benefits of the K-means algorithm in deploying the WSN.

Biometric traits gradually proved their importance in real-life applications, especially in identification field. Among the available biometric traits, the unique shape of the human ear has also received loads of attention from scientists through the years. Hence, numerous ear-based approaches have been proposed with promising performance. With these methods, plenty problems can be solve by the distinctiveness of ear features, such as recognizing human with mask or diagnose ear-related diseases. As a complete identification system requires an effective detector for real-time application, and the current richness and variety of ear detection algorithms are poor due to the small and complex shape of human ears. In this paper, we introduce a new human ear detection pipeline based on the YOLOv3 detector. A well-known face detector named RetinaFace is also added in the detection system to narrow the regions of interest and enhance the accuracy. The proposed method is evaluated on an unconstrained dataset, which shows its effectiveness.

Summary: RAS proteins function as highly regulated molecular switches that control cellular growth. In addition to regulatory proteins, RAS undergoes a number of posttranslational modifications (PTMs) that regulate its activity. Lysine 104, a hot spot for multiple PTMs, is a highly conserved residue that forms key interactions that stabilize the RAS helix-2(H2)/helix-3(H3) interface. Mutation at 104 attenuates interaction with guanine nucleotide exchange factors (GEFs), whereas ubiquitination at lysine 104 retains GEF regulation. To elucidate how ubiquitination modulates RAS function, we generated monoubiquitinated KRAS at 104 using chemical biology approaches and conducted biochemical, NMR, and computational analyses. We find that ubiquitination promotes a new dynamic interaction network and alters RAS conformational dynamics to retain GEF function. These findings reveal a mechanism by which ubiquitination can regulate protein function.

Gesture recognition technology is rapidly growing in the recent years due to the demands of many application such as computer game and sport, human robot interaction, assistant systems, sign language interpretation and e-commerce. One of the most important of gesture recognition is hand-gesture recognition. For example, it can be used to control all devices (television, radio, air-condition, and doors) by just hand gestures for smart home application. The HGM-4 dataset is built for hand gesture recognition (the full dataset is available from: https://data.mendeley.com/datasets/jzy8zngkbg/4) which contains total 4,160 color images (1280 × 700 pixels) of 26 hand gestures captured by four cameras at different position. The training and testing set are defined to create a benchmark framework for comparing the experimental results.

Forest carbon sequestration offsets are methodologically uncertain, comprise a minor component of carbon markets and do not effectively slow deforestation. The objective of this study is to describe a commercial scale in situ measurement approach for determination of net forest carbon sequestration projects, the Direct Measurement Forest Carbon Protocol™, to address forest carbon market uncertainties. In contrast to protocols that rely on limited forest mensuration, growth simulation and exclusion of CO2 data, the Direct Measurement Forest Carbon Protocol™ is based on standardized methods for direct determination of net ecosystem exchange (NEE) of CO2 employing eddy covariance, a meteorological approach integrating forest carbon fluxes. NEE is used here as the basis for quantifying the first of its kind carbon financial products. The DMFCP differentiates physical, project and financial carbon within a System-of-Systems™ (SoS) network architecture. SoS sensor nodes, the Global Monitoring Platform™ (GMP), housing analyzers for CO2 isotopologues (e.g., 12CO2,13CO2, 14CO2) and greenhouse gases are deployed across the project landscape. The SoS standardizes and automates GMP measurement, uncertainty and reporting functions creating diverse forest carbon portfolios while reducing cost and investment risk in alignment with modern portfolio theory. To illustrate SoS field deployment and operation, published annual NEE data for a tropical (Ankasa Park, Ghana, Africa) and a deciduous forest (Harvard Forest, Petersham, MA, USA) are used to forecast carbon revenue. Carbon pricing scenarios are combined with historical in situ NEE annual time-series to extrapolate pre-tax revenue for each project applied to 100,000 acres (40,469 hectares) of surrounding land. Based on carbon pricing of $5 to $36 per ton CO2 equivalent (tCO2eq) and observed NEE sequestration rates of 0.48 to 15.60 tCO2eq acre−1 yr−1, pre-tax cash flows ranging from $230,000 to $16,380,000 across project time-series are calculated, up to 5× revenue for contemporary voluntary offsets, demonstrating new economic incentives to reverse deforestation. The SoS concept of operation and architecture, with engineering development, can be extended to diverse gas species across terrestrial, aquatic and oceanic ecosystems, harmonizing voluntary and compliance market products worldwide to assist in the management of global warming. The Direct Measurement Forest Carbon Protocol reduces risk of invalidation intrinsic to estimation-based protocols such as the Climate Action Reserve and the Clean Development Mechanism that do not observe molecular CO2 to calibrate financial products. Multinational policy applications such as the Paris Agreement and the United Nations Reducing Emissions from Deforestation and Degradation, constrained by Kyoto Protocol era processes, will benefit from NEE measurement avoiding unsupported claims of emission reduction, fraud, and forest conservation policy failure.

Ear recognition is starting to grow as an alternative to other biometric recognition types in recent years. The EarVN1.0 dataset is constructed by collecting ear images of 164 Asian peoples during 2018. It is among the largest ear datasets publicly to the research community which composed by 28,412 colour images of 98 males and 66 females. Thus, this dataset is different from previous works by providing images of both ears per person under unconstrained conditions. The original facial images have been acquired by unconstrained environment including cameras systems and light condition. Ear images are then cropped from facial images over the large variations of pose, scale and illumination. Several machine learning tasks can be applied such as ear recognition, image classification or clustering, gender recognition, right-ear or left-ear detection and enhanced super resolution. Keywords: Ear recognition, Biometric, Image classification, Super-resolution, Image clustering, Right-ear or left-ear detection

LBP (Local Binary Pattern) is a very popular texture descriptor largely used in computer vision. In most applications, LBP histograms are exploited as texture features leading to a high dimensional feature space, especially for color texture classification problems. In the past few years, different solutions were proposed to reduce the dimension of the feature space based on the LBP histogram. Most of these approaches apply feature selection methods in order to find the most discriminative bins. Recently another strategy proposed selecting the most discriminant LBP histograms in their entirety. This paper tends to improve on these previous approaches, and presents a combination of LBP bin and histogram selections, where a histogram ranking method is applied before processing a bin selection procedure. The proposed approach is evaluated on five benchmark image databases and the obtained results show the effectiveness of the combination of LBP bin and histogram selections which outperforms the simple LBP bin and LBP histogram selection approaches when they are applied independently.

Tuberculosis (TB) causes more deaths worldwide than any other infectious disease. Progress in reducing the global burden of TB stalled during the COVID-19 pandemic; an estimated 10.6 million persons became [...]

BACKGROUND:DNA methylation is affected by the activities of the key enzymes and intermediate metabolites of the one-carbon pathway, one of which involves homocysteine. We investigated the effect of the well-known genetic variant associated with mildly elevated homocysteine: MTHFR 677C>T independently and in combination with other homocysteine-associated variants, on genome-wide leukocyte DNA-methylation. METHODS:Methylation levels were assessed using Illumina 450k arrays on 9,894 individuals of European ancestry from 12 cohort studies. Linear-mixed-models were used to study the association of additive MTHFR 677C>T and genetic-risk score (GRS) based on 18 homocysteine-associated SNPs, with genome-wide methylation. RESULTS:Meta-analysis revealed that the MTHFR 677C>T variant was associated with 35 CpG sites in cis, and the GRS showed association with 113 CpG sites near the homocysteine-associated variants. Genome-wide analysis revealed that the MTHFR 677C>T variant was associated with 1 trans-CpG (nearest gene ZNF184), while the GRS model showed association with 5 significant trans-CpGs annotated to nearest genes PTF1A, MRPL55, CTDSP2, CRYM and FKBP5. CONCLUSIONS:Our results do not show widespread changes in DNA-methylation across the genome, and therefore do not support the hypothesis that mildly elevated homocysteine is associated with widespread methylation changes in leukocytes.

Bariatric surgery is associated to improvements in obesity-associated comorbidities thought to be mediated by a decrease of adipose inflammation. However, the molecular mechanisms behind these beneficial effects are poorly understood.We analyzed RNA-seq expression profiles in adipose tissue from 22 obese women before and 3 months after surgery. Of 15,972 detected genes, 1214 were differentially expressed after surgery at a 5% false discovery rate. Upregulated genes were mostly involved in the basal cellular machinery. Downregulated genes were enriched in metabolic functions of adipose tissue. At baseline, 26 modules of coexpressed genes were identified. The four most stable modules reflected the innate and adaptive immune responses of adipose tissue. A first module reflecting a non-specific signature of innate immune cells, mainly macrophages, was highly conserved after surgery with the exception of DUSP2 and CD300C. A second module reflected the adaptive immune response elicited by T lymphocytes; after surgery, a disconnection was observed between genes involved in T-cell signaling and mediators of the signal transduction such as CXCR1, CXCR2, GPR97, CCR7 and IL7R. A third module reflected neutrophil-mediated inflammation; after surgery, several genes were dissociated from the module, including S100A8, S100A12, CD300E, VNN2, TUBB1 and FAM65B. We also identified a dense network of 19 genes involved in the interferon-signaling pathway which was strongly preserved after surgery, with the exception of DDX60, an antiviral factor involved in RIG-I-mediated interferon signaling. A similar loss of connection was observed in lean mice compared to their obese counterparts.These results suggest that improvements of the inflammatory state following surgery might be explained by a disruption of immuno-inflammatory cascades involving a few crucial molecules which could serve as potential therapeutic targets.

Trong số 1.652 chó đến khám và điều trị lần đầu tại Bệnh xá Thú Y và các Phòng mạch Thú y Thành phố Cần Thơ, chúng tôi phát hiện 122 chó có bệnh lý ở hệ tiết niệu. Qua kết quả kiểm tra X-quang và xét nghiệm nước tiểu, xác định chó có bệnh lý trên hệ tiết niệu chiếm tỷ lệ 7,38%. Trong số đó, chó có bệnh lý suy thận mãn xuất hiện với tỷ lệ cao nhất 31,97%, kế đến là suy thận cấp 28,69%, bệnh lý viêm nhiễm trùng bàng quang 22,95%, sỏi bàng quang và niệu đạo 16,39%. Bệnh lý trên hệ niệu của chó khác biệt có ý nghĩa thống kê về độ tuổi, giống chó với p

In order to investigate whether DNA methylation marks could contribute to the incomplete penetrance of the FV Leiden mutation, a major genetic risk factor for venous thrombosis (VT), we measured genome-wide DNA methylation levels in peripheral blood samples of 98 VT patients carrying the mutation and 251 VT patients without the mutation using the dedicated Illumina HumanMethylation450 array. The genome-wide analysis of 388,120 CpG probes identified three sites mapping to the SLC19A2 locus whose DNA methylation levels differed significantly (p0.05). In conclusion, our work clearly illustrates some promises and pitfalls of DNA methylation investigations on peripheral blood DNA in large epidemiological cohorts. DNA methylation levels at SLC19A2 are influenced by SNPs in LD with FV Leiden, but these DNA methylation marks do not explain the incomplete penetrance of the FV Leiden mutation.

In order to assess whether gene expression variability could be influenced by several SNPs acting in cis, either through additive or more complex haplotype effects, a systematic genome-wide search for cis haplotype expression quantitative trait loci (eQTL) was conducted in a sample of 758 individuals, part of the Cardiogenics Transcriptomic Study, for which genome-wide monocyte expression and GWAS data were available. 19,805 RNA probes were assessed for cis haplotypic regulation through investigation of ~2,1 × 10(9) haplotypic combinations. 2,650 probes demonstrated haplotypic p-values >10(4)-fold smaller than the best single SNP p-value. Replication of significant haplotype effects were tested for 412 probes for which SNPs (or proxies) that defined the detected haplotypes were available in the Gutenberg Health Study composed of 1,374 individuals. At the Bonferroni correction level of 1.2 × 10(-4) (~0.05/412), 193 haplotypic signals replicated. 1000 G imputation was then conducted, and 105 haplotypic signals still remained more informative than imputed SNPs. In-depth analysis of these 105 cis eQTL revealed that at 76 loci genetic associations were compatible with additive effects of several SNPs, while for the 29 remaining regions data could be compatible with a more complex haplotypic pattern. As 24 of the 105 cis eQTL have previously been reported to be disease-associated loci, this work highlights the need for conducting haplotype-based and 1000 G imputed cis eQTL analysis before commencing functional studies at disease-associated loci.

BackgroundThe hypothesis of dosage compensation of genes of the X chromosome, supported by previous microarray studies, was recently challenged by RNA-sequencing data. It was suggested that microarray studies were biased toward an over-estimation of X-linked expression levels as a consequence of the filtering of genes below the detection threshold of microarrays.Methodology/principal findingsTo investigate this hypothesis, we used microarray expression data from circulating monocytes in 1,467 individuals. In total, 25,349 and 1,156 probes were unambiguously assigned to autosomes and the X chromosome, respectively. Globally, there was a clear shift of X-linked expressions toward lower levels than autosomes. We compared the ratio of expression levels of X-linked to autosomal transcripts (X∶AA) using two different filtering methods: 1. gene expressions were filtered out using a detection threshold irrespective of gene chromosomal location (the standard method in microarrays); 2. equal proportions of genes were filtered out separately on the X and on autosomes. For a wide range of filtering proportions, the X∶AA ratio estimated with the first method was not significantly different from 1, the value expected if dosage compensation was achieved, whereas it was significantly lower than 1 with the second method, leading to the rejection of the hypothesis of dosage compensation. We further showed in simulated data that the choice of the most appropriate method was dependent on biological assumptions regarding the proportion of actively expressed genes on the X chromosome comparative to the autosomes and the extent of dosage compensation.Conclusion/significanceThis study shows that the method used for filtering out lowly expressed genes in microarrays may have a major impact according to the hypothesis investigated. The hypothesis of dosage compensation of X-linked genes cannot be firmly accepted or rejected using microarray-based data.