Design of an Intelligent Approach on Capsule Networks to Detect Forged Images

Forgeries have recently become more prevalent in the society as a result of recent improvements in media generation technologies. In real-time, modern technology allows for the creation of a forged version of a single image obtained from a social network. Forgery detection algorithms have been created for a variety of areas; however they quickly become obsolete as new attack types exist. This paper presents a unique image forgery detection strategy based on deep learning algorithms. The proposed approach employs a convolutional neural network (CNN) to produce histogram representations from input RGB color images, which are then utilized to detect image forgeries. With the image separation method and copy-move detection applications in mind, the proposed CNN is combined with an intelligent approach and histogram mapping. It is used to detect fake or true images at the initial stage of our proposed work. Besides, it is specially designed for performing feature extraction in image layer separation with the help of CNN model. To capture both geographical and histogram information and the likelihood of presence at the same time, we use vectors in our dynamic capsule networks to detect the forgery kernels from reference images. The proposed research work integrates the intelligence with a feature engineering approach in an efficient manner. They are well-known and efficient in the identification of forged images. The performance metrics such as accuracy, recall, precision, and half total error rate (HTER) are computed and tabulated with the graph plot.

Long Range PCR-based deep sequencing for haplotype determination in mixed HCMV infections

Background Short read sequencing, which has extensively been used to decipher the genome diversity of human cytomegalovirus (HCMV) strains, often falls short to assess co-linearity of non-adjacent polymorphic sites in mixed HCMV populations. In the present study, we established a long amplicon sequencing workflow to identify number and relative quantities of unique HCMV haplotypes in mixtures. Accordingly, long read PacBio sequencing was applied to amplicons spanning over multiple polymorphic sites. Initial validation of this approach was performed with defined HCMV DNA templates derived from cell-free viruses and was further tested for its suitability on patient samples carrying mixed HCMV infections. Results Our data show that artificial HCMV DNA mixtures were correctly determined upon long amplicon sequencing down to 1% abundance of the minor DNA source. Total error rate of mapped reads ranged from 0.17 to 0.43 depending on the stringency of quality trimming. PCR products of up to 7.7 kb and a GC content < 55% were efficiently generated when DNA was directly isolated from bronchoalveolar lavage samples, yet long range PCR may display a slightly lower sensitivity compared to short amplicons. In a single sample, up to three distinct haplotypes were identified showing varying relative frequencies. Intra-patient haplotype diversity is unevenly distributed across the target site and often interspersed by long identical stretches, thus unable to be linked by short reads. Moreover, diversity at single polymorphic regions as assessed by short amplicon sequencing may markedly underestimate the overall diversity of mixed populations. Conclusions Quantitative haplotype determination by long amplicon sequencing provides a novel approach for HCMV strain characterisation in mixed infected samples which can be scaled up to cover the majority of the genome. This will substantially improve our understanding of intra-host HCMV strain diversity and its dynamic behaviour.

Enhancement of Spectrum Sensing Performance via Cooperative Cognitive Radio ‎Networks at Low SNR

The inefficient use of spectrum is the key subject to overcome the upcoming spectrum crunch issue. This paper presents a study of performance of cooperative cognitive network via hard combining of decision fusion schemes. Simulation results presented different cooperative hard decision fusion schemes for cognitive network. The hard-decision fusion schemes provided different discriminations for detection levels. They also produced small values of Miss-Detection Probability at different values of Probability of False Alarm and adaptive threshold levels. The sensing performance was investigated under the influence of channel condition for proper operating conditions. An increase in the detection performance was achieved for cognitive users (secondary users) of the authorized unused dynamic spectrum holes (primary users) while operating in a very low signal-to-noise ratio with the proper condition of minimum total error rate.

Penerapan dan Perbandingan Tiga Metode Analisis Pohon Keputusan pada Klasifikasi Penderita Kanker Payudara

Today there is a considerable amount of work dealing with decision trees, especially in survival analysis (Ibrahim et al, 2008). Cases classified as survival analysis, like cancer patients. This study discusses the application of data mining which is to obtain diagnostic results. The classification technique uses information obtained from medical records of breast cancer patients in Yugoslavia. A method for answering these problems through decision tree analysis using the CHAID, Exhaustive CHAID and CART methods. Empirically aiming to compare performance of three decision tree classification methods so that the best method is obtained. It was concluded that best method used in applying to the classification of breast cancer sufferers was the CART method because it was able to get the most significant variables at most four, namely inv-node, tumor size, deg-malig and breast parts. Then it has a total accuracy rate with highest value of 84.9 percent and has a total error rate with lowest value of 15.1 percent. Abstrak. Dewasa ini ada cukup banyak pekerjaan yang berurusan dengan pohon keputusan, terutama dalam analisis survival (Ibrahim dkk, 2008). Kasus yang tergolong analisis survival seperti penderita penyakit kanker. Penelitian ini membahas mengenai penerapan data mining yang digunakan untuk mendapatkan hasil diagnostik. Pendekatan teknik klasifikasi dengan menggunakan informasi yang diperoleh pada rekam medis data penderita kanker payudara di Yugoslavia. Salah satu metode untuk menjawab permasalahan tersebut melalui analisis pohon keputusan dengan metode CHAID, Exhaustive CHAID dan CART. Secara empiris bertujuan untuk membandingkan kinerja tiga metode pengklasifikasi pohon keputusan agar didapatkan metode manakah yang terbaik. Maka disimpulkan bahwa metode terbaik yang digunakan dalam penerapan pada klasifikasi penderita kanker payudara adalah metode CART sebab mampu mendapatkan variabel signifikan yang paling banyak ada empat, yakni inv-node, ukuran tumor, deg-malig dan bagian payudara. Kemudian memiliki tingkat akurasi total dengan nilai tertinggi sebesar 84.9 persen dan memiliki total tingkat kesalahan dengan nilai yang terendah sebesar 15.1 persen.

Long Range PCR-based deep sequencing for haplotype determination in mixed HCMV infections

Short read sequencing, which has extensively been used to decipher the genome diversity of human cytomegalovirus (HCMV) strains, often falls short to assess co-linearity of non-adjacent polymorphic sites in mixed HCMV populations. In the present study, we established a long amplicon sequencing workflow to identify number and relative quantities of unique HCMV haplotypes in mixtures. Accordingly, long read PacBio sequencing was applied to amplicons spanning over multiple polymorphic sites. Initial validation of this approach was performed with defined HCMV DNA templates derived from cell-free viruses and was further tested for its suitability on patient samples carrying mixed HCMV infections. Our data show that artificial HCMV DNA mixtures were correctly determined upon long amplicon sequencing down to 1% abundance of the minor DNA source. Total error rate of mapped reads ranged from 0.17 to 0.43 depending on the stringency of quality trimming. PCR products of up to 7.7 kb and a GC content <55% were efficiently generated when DNA was directly isolated from bronchoalveolar lavage samples, yet long range PCR may display a slightly lower sensitivity compared to short amplicons. In a single sample, up to three distinct haplotypes were identified showing varying relative frequencies. Intra-patient haplotype diversity is unevenly distributed across the target site and often interspersed by long identical stretches, thus unable to be linked by short reads. Moreover, diversity at single polymorphic regions as assessed by short amplicon sequencing may markedly underestimate the overall diversity of mixed populations. Quantitative haplotype determination by long amplicon sequencing provides a novel approach for HCMV strain characterisation in mixed infected samples which can be scaled up to cover the majority of the genome. This will substantially improve our understanding of intra-host HCMV strain diversity and its dynamic behaviour.

Comparing the efficiency of paper-based and electronic data capture during face-to-face interviews

On-site surveys involving face-to-face interviews are implemented globally across many scientific disciplines. Incorporating new technologies into such surveys by using electronic devices is becoming more common and is widely viewed to be more cost-effective and accurate. However, Electronic Data Capture methods (EDC) when compared to traditional Paper-based Data Capture (PDC) are often implemented without proper evaluation of any changes in efficiency, especially from surveys in coastal and marine environments. A roving creel survey of recreational shore-based fishers in Western Australia in 2019 enabled a direct comparison between the two methods. Randomisation strategies were employed to ensure biases in using each technique were minimised. A total of 1,068 interviews with recreational fishers were undertaken with a total error rate of 5.1% (CI95%: 4.8–5.3%) for PDC and 3.1% (CI95%: 2.9–3.3%) for EDC. These results confirmed that EDC can reduce errors whilst increasing efficiency and decreasing cost, although some aspects of this platform could be improved with some streamlining. This study demonstrates how EDC can be successfully implemented in coastal and marine environments without compromising the randomised, stratified nature of a survey and highlights the cost-effectiveness of this method. Such findings can be widely applied to any discipline which uses face-to-face interviews for data collection.

An analysis of energy detector based on improved approximations of the chi-square distributions

Closed-form expressions for the detection probability, the false alarm probability and the energy detector constant threshold are derived using approximations of the central chi-square and non-central chi-square distributions. The approximations used show closer proximity to the original functions when compared to the expressions used in the literature. The novel expressions allow gains up to 6% and 16% in terms of measured false alarm and miss-detection probability, respectively, if compared to the Central Limit Theorem approach. The throughput of cognitive network is also enhanced when these novel expressions are implemented, providing gains up to 9%. New equations are also presented that minimize the total error rate to obtain the detection threshold and the optimal number of samples. The analytical results match the results of the simulation for a wide range of SNR values.

Video Forensics: Identifying Colorized Images Using Deep Learning

In recent years there has been a significant increase in images and videos circulating in social networks and media, edited with different techniques, including colorization. This has a negative impact on the forensic field because it is increasingly difficult to discern what is original content and what is fake. To address this problem, we propose two models (a custom architecture and a transfer-learning-based model) based on CNNs that allows a fast recognition of the colorized images (or videos). In the experimental test, the effect of three hyperparameters on the performance of the classifier were analyzed in terms of HTER (Half Total Error Rate). The best result was found for the Adam optimizer, with a dropout of 0.25 and an input image size of 400 × 400 pixels. Additionally, the proposed models are compared with each other in terms of performance and inference times and with some state-of-the-art approaches. In terms of inference times per image, the proposed custom model is 12x faster than the transfer-learning-based model; however, in terms of precision (P), recall and F1-score, the transfer-learning-based model is better than the custom model. Both models generalize better than other models reported in the literature.

Is There a Minimum Number of Landmarks That Optimizes the Geometric Morphometric Analysis of Mosquito (Diptera, Culicidae) Wings?

Culicids are the most significant arthropods affecting human health. Thus, their correct identification is critical. The use of Geometric Morphometrics (GM) has been recently incorporated into mosquito taxonomy and has begun to complement classic diagnostic techniques. Since sampling size depends on the number of Landmarks (LMs) used, this study aimed to establish the minimum number of wing LMs needed to optimize GM analysis of mosquito species and/or genera from urban and peri-urban areas of Argentina. Female left wings were used for the optimization phase, in which 17 LMs were reduced to four by iterative LM exclusion. To verify its efficiency, Principal Component Analysis (PCA), Discriminant Analysis (DA), and Canonical Variate Analysis (CVA) were performed. Additionally, a phenogram was constructed to visualize the results. We observed that five LMs for the PCA, CVA, and phenogram and nine for the DA enabled discrimination and/or clustering of almost all species and genera. Therefore, we tested the LM selection by using nine LMs and adding new species. The resulting PCA showed little overlap between species and almost all species clustered as expected, which was also reflected in the phenogram. Significant differences were found between wing shape among all species, together with a low total error rate in the DA. In conclusion, the number of LMs can be reduced and still be used to effectively differentiate and cluster culicids. This is helpful for better exploitation of available material and optimization of data processing time when classic taxonomy methods are inadequate or the material is scarce.