Search Results (61)

Counterfeit products cause financial losses for both the manufacturer and the enduser; e.g., fake foods and medicines pose significant risks to the public’s health. Moreover, it is challenging to ensure trust in a product’s supply chain, preventing counterfeit goods from being distributed throughout the network. However, fake product detection methods are expensive and need to be more scalable, whereas a unified traceability system for packaged products is not available. Therefore, this research proposes a product traceability system, named Trusted Traceability Service (TTS), using Blockchain and Self-Sovereign Identity (SSI). The TTS can be incorporated across diverse industries because of its generic and manageable four-layer product packaging strategy. Blockchain-enabled SSI empowers distributed nodes, to verify them without a centralized client–server authorization architecture. Moreover, due to its distributed nature, the proposed TTS framework is scalable and robust, with the use of web3.0 distributed application development. The adoption of Fantom, a public blockchain infrastructure, allows the proposed system to handle thousands of successful transactions more cost-effectively than the Ethereum network. The deployment of the proposed framework in both public and private blockchain networks demonstrated its superiority in execution time and number of successful transactions. Full article

Currently, the phenomenon of deepfakes is becoming increasingly significant, as they enable the creation of extremely realistic images capable of deceiving anyone thanks to deep learning tools based on generative adversarial networks (GANs). These images are used as profile pictures on social media with the intent to sow discord and perpetrate scams on a global scale. In this study, we demonstrate that these images can be identified through various imperfections present in the synthesized eyes, such as the irregular shape of the pupil and the difference between the corneal reflections of the two eyes. These defects result from the absence of physical and physiological constraints in most GAN models. We develop a two-level architecture capable of detecting these fake images. This approach begins with an automatic segmentation method for the pupils to verify their shape, as real image pupils naturally have a regular shape, typically round. Next, for all images where the pupils are not regular, the entire image is analyzed to verify the reflections. This step involves passing the facial image through an architecture that extracts and compares the specular reflections of the corneas of the two eyes, assuming that the eyes of real people observing a light source should reflect the same thing. Our experiments with a large dataset of real images from the Flickr-FacesHQ and CelebA datasets, as well as fake images from StyleGAN2 and ProGAN, show the effectiveness of our method. Our experimental results on the Flickr-Faces-HQ (FFHQ) dataset and images generated by StyleGAN2 demonstrated that our algorithm achieved a remarkable detection accuracy of 0.968 and a sensitivity of 0.911. Additionally, the method had a specificity of 0.907 and a precision of 0.90 for this same dataset. And our experimental results on the CelebA dataset and images generated by ProGAN also demonstrated that our algorithm achieved a detection accuracy of 0.870 and a sensitivity of 0.901. Moreover, the method had a specificity of 0.807 and a precision of 0.88 for this same dataset. Our approach maintains good stability of physiological properties during deep learning, making it as robust as some single-class deepfake detection methods. The results of the tests on the selected datasets demonstrate higher accuracy compared to other methods. Full article

Breast cancer ranks as the second most prevalent cancer globally and is the most frequently diagnosed cancer among women; therefore, early, automated, and precise detection is essential. Most AI-based techniques for breast cancer detection are complex and have high computational costs. Hence, to overcome this challenge, we have presented the innovative LightweightUNet hybrid deep learning (DL) classifier for the accurate classification of breast cancer. The proposed model boasts a low computational cost due to its smaller number of layers in its architecture, and its adaptive nature stems from its use of depth-wise separable convolution. We have employed a multimodal approach to validate the model’s performance, using 13,000 images from two distinct modalities: mammogram imaging (MGI) and ultrasound imaging (USI). We collected the multimodal imaging datasets from seven different sources, including the benchmark datasets DDSM, MIAS, INbreast, BrEaST, BUSI, Thammasat, and HMSS. Since the datasets are from various sources, we have resized them to the uniform size of 256 × 256 pixels and normalized them using the Box-Cox transformation technique. Since the USI dataset is smaller, we have applied the StyleGAN3 model to generate 10,000 synthetic ultrasound images. In this work, we have performed two separate experiments: the first on a real dataset without augmentation and the second on a real + GAN-augmented dataset using our proposed method. During the experiments, we used a 5-fold cross-validation method, and our proposed model obtained good results on the real dataset (87.16% precision, 86.87% recall, 86.84% F1-score, and 86.87% accuracy) without adding any extra data. Similarly, the second experiment provides better performance on the real + GAN-augmented dataset (96.36% precision, 96.35% recall, 96.35% F1-score, and 96.35% accuracy). This multimodal approach, which utilizes LightweightUNet, enhances the performance by 9.20% in precision, 9.48% in recall, 9.51% in F1-score, and a 9.48% increase in accuracy on the combined dataset. The LightweightUNet model we proposed works very well thanks to a creative network design, adding fake images to the data, and a multimodal training method. These results show that the model has a lot of potential for use in clinical settings. Full article

Nowadays, the consequence of quick access to information has lead to the spread of fake news, which has a strong damaging impact on democracy, justice, and public trust. Thus, it is crucial to analyze and evaluate detection methods for fake news. This paper focuses on the detection of Romanian fake news. In this study, we made a comparative analysis of machine learning algorithms and Transformer-based models on Romanian fake news detection using three datasets—FakeRom, NEW, and both FakeRom + NEW. The NEW dataset was build using a scrapping algorithm applied on the Veridica platform. Our approach uses the following machine learning models for detection: Naive Bayes (NB), Logistic Regression (LR), and Support Vector Machine (SVM). We also used two Transformer-based models—BERT-based-multilingual-cased and RoBERTa-large. The performance of the models was evaluated using various metrics: accuracy, precision, recall, and F1 score. The results revealed that the BERT model trained on the NEW dataset consistently achieved the highest performance metrics across all test sets, with 96.5%. Also, Support Vector Machine trained on NEW was another top performer, reaching a very good accuracy of 94.6% on the combined test set. Full article

We propose a semi-supervised masked auxiliary classifier generative adversarial network (SM-ACGAN) that has good classification performance in situations where labeled training data are limited. To develop SM-ACGAN, we combine the strengths of SSGAN (semi-supervised GAN), ACGAN-SG (auxiliary classifier GAN based on spectral normalization and gradient penalty), and MaskedGAN. Additionally, we devise a novel masking technique that performs masking adaptively depending on the real/fake ratio of the input data and a novel regularization technique that stabilizes the generator training depending on the maximum ratio of the average power of the generated fake data to the average power of the noise latent variables. Finally, we devise a rule of selecting an appropriate quantity of unlabeled data and labeled fake data generated by the generator for effective data augmentation. Through simulations, we demonstrate that SM-ACGAN has lower root mean square error (RMSE) values and lower variance, demonstrating superior mobile speed measurement performance on Rician channels compared to ACGAN-SG, MaskedGAN, SSGAN, a CNN (convolutional neural network), and a DNN (deep neural network). Full article

Deepfake has become an emerging technology affecting cyber-security with its illegal applications in recent years. Most deepfake detectors utilize CNN-based models such as the Xception Network to distinguish real or fake media; however, their performance on cross-datasets is not ideal because they suffer from over-fitting in the current stage. Therefore, this paper proposed a spatial consistency learning method to relieve this issue in three aspects. Firstly, we increased the selections of data augmentation methods to 5, which is more than our previous study’s data augmentation methods. Specifically, we captured several equal video frames of one video and randomly selected five different data augmentations to obtain different data views to enrich the input variety. Secondly, we chose Swin Transformer as the feature extractor instead of a CNN-based backbone, which means that our approach did not utilize it for downstream tasks, and could encode these data using an end-to-end Swin Transformer, aiming to learn the correlation between different image patches. Finally, this was combined with consistency learning in our study, and consistency learning was able to determine more data relationships than supervised classification. We explored the consistency of video frames’ features by calculating their cosine distance and applied traditional cross-entropy loss to regulate this classification loss. Extensive in-dataset and cross-dataset experiments demonstrated that Swin-Fake could produce relatively good results on some open-source deepfake datasets, including FaceForensics++, DFDC, Celeb-DF and FaceShifter. By comparing our model with several benchmark models, our approach shows relatively strong robustness in detecting deepfake media. Full article

With the development of internet technology, fake news has become a multi-modal collection. The current news detection methods cannot fully extract semantic information between modalities and ignore the rumor properties of fake news, making it difficult to achieve good results. To address the problem of the accurate identification of multi-modal fake news, we propose the Vae-Clip multi-modal fake news detection model. The model uses the Clip pre-trained model to jointly extract semantic features of image and text information using text information as the supervisory signal, solving the problem of semantic interaction across modalities. Moreover, considering the rumor attributes of fake news, we propose to fuse semantic features with rumor style features using multi-feature fusion to improve the generalization performance of the model. We use a variational autoencoder to extract rumor style features and combine semantic features and rumor features using an attention mechanism to detect fake news. Numerous experiments were conducted on four datasets primarily composed of Weibo and Twitter, and the results show that the proposed model can accurately identify fake news and is suitable for news detection in complex scenarios, with the highest accuracy reaching 96.3%. Full article

The main purpose of this study is to generate defect images of body parts using a GAN (generative adversarial network) and compare and analyze the performance of the YOLO (You Only Look Once) v7 and v8 object detection models. The goal is to accurately judge good and defective products. Quality control is very important in the automobile industry, and defects in body parts directly affect vehicle safety, so the development of highly accurate defect detection technology is essential. This study ensures data diversity by generating defect images of car body parts using a GAN and through this, compares and analyzes the object detection performance of the YOLO v7 and v8 models to present an optimal solution for detecting defects in car parts. Through experiments, the dataset was expanded by adding fake defect images generated by the GAN. The performance experiments of the YOLO v7 and v8 models based on the data obtained through this approach demonstrated that YOLO v8 effectively identifies objects even with a smaller amount of data. It was confirmed that defects could be detected. The readout of the detection system can be improved through software calibration. Full article

Fingermarks play an important role in document identification. At the same time, fingermarks on paper documents are often accompanied by signatures and background text, which introduce noise to the original fingermark textures and increase the difficulty of detection. A signed fingermark detection method based on deep residual networks and a decision-level fusion strategy was proposed to defend against spoofing attacks from fake fingermarks. Firstly, the multi-scale structure was introduced in the residual module, which improved the network’s depth and breadth without increasing the parameters. Then, the multi-probability label strategy was refined and employed to enhance the local encoding ability of the feature extraction. A score fusion strategy was designed, with weights allocated based on the difference in signed interference levels of local image blocks. Finally, a model fusion strategy based on evidence theory was suggested, which improved detection accuracy by leveraging complementarity between models. A large-scale fingermark database was established, which included real fingermarks made from real fingers and fake fingermarks made from various materials, and this was divided into two sub databases: signed and unsigned. The experimental results show that the proposed method achieves 96.16% accuracy based on the fingerprint dataset of the global liveness detection competition called LivDet2017 and achieves 99.30% accuracy based on the signed fingermark database, while it has good resistance to spoofing attacks from unknown materials. Full article

The issue of the dissemination of fake news has been widely addressed in the literature, but the issue of the dissemination of fact checks to debunk fake news has not received sufficient attention. Fake news is tailored to reach a wide audience, a concern that, as this paper shows, does not seem to be present in fact checking. As a result, fact checking, no matter how good it is, fails in its goal of debunking fake news for the general public. This paper addresses this problem with the aim of increasing the effectiveness of the fact checking of online social media posts through the use of cognitive tools, yet grounded in ethical principles. The paper consists of a profile of the prevalence of fact checking in online social media (both from the literature and from field data) and an assessment of the extent to which engagement can be increased by using simple cognitive enhancements in the text of the post. The focus is on Snopes and $\mathbb{X}$ (formerly Twitter). Full article

We compared three methods for scoring the Balanced Inventory of Desirable Responding (BIDR) to detect faked responses on self-report measures: (1) polytomous, (2) dichotomous emphasizing exaggerating endorsement of socially desirable behaviors, and (3) dichotomous emphasizing exaggerating denial of such behaviors. The results revealed that respondents on average were able to fake good or fake bad and that faking markedly affected score distributions, subscale score intercorrelations, and overall model fits. When using the Impression Management scale, polytomous and dichotomous exaggerated endorsement scoring were best for detecting faking good, whereas polytomous and dichotomous exaggerated denial scoring were best for detecting faking bad. When using the Self-Deceptive Enhancement scale, polytomous and dichotomous exaggerated endorsement scoring again were best for detecting faking good, but dichotomous exaggerated denial scoring was best for detecting faking bad. Percentages of correct classification of honest and faked responses for the most effective methods for any given scale ranged from 85% to 93%, with accuracy on average in detecting faking bad greater than in detecting faking good and greater when using the Impression Management than using the Self-Deceptive Enhancement scale for both types of faking. Overall, these results best support polytomous scoring of the BIDR Impression Management scale as the single most practical and efficient means to detect faking. Cut scores that maximized classification accuracy for all scales and scoring methods are provided for future use in screening for possible faking within situations in which relevant local data are unavailable. Full article

Several forged 3-Polker coins have been reported in historical sources on the financial crisis that occurred between 1619 and 1623 at the start of the 30-year-long war. Supposedly, belligerent countries forged other countries’ coins which were then used for external payments as a war strategy. Thus, a lot of 3-Polker coins (e.g., Sigismund-III-type) were forged, and the markets became flooded with poor currency. In the present day, these pre-modern forgeries are rare archeological findings. Only five forged 3-Polker coins randomly found in Transylvania were available for the current study. There are deeper implications of silver and tin in the forgery techniques that need to be considered. Thus, the forged 3-Polker coins were investigated via nondestructive methods: SEM microscopy coupled with EDS elemental spectroscopy for complex microstructural characterization and XRD for phase identification. Three distinct types of forgery methods were identified: the amalgam method is the first used for copper blank silvering (1620), and immersion in melted silver (1621) is the second one. Both methods were used to forge coins with proper legends and inscriptions. The third method is the tin plating of copper coins (with corrupted legend and altered design) (1622, 1623, and 1624). The EDS investigation revealed Hg traces inside the compact silver crusts for the first type and the elongated silver crystallites in the immersion direction, which are well-attached to the copper core for the second type. The third forgery type has a rich tin plating with the superficial formation of Cu₆Sn₅ compound that assures a good resistance of the coating layer. Therefore, this type should have been easily recognized as fake by traders, while the first two types require proper weighing and margin clipping to ensure their quality. Full article

With the rapid development of fingerprint recognition systems, fingerprint liveness detection is gradually becoming regarded as the main countermeasure to protect the fingerprint identification system from spoofing attacks. Convolutional neural networks have shown great potential in fingerprint liveness detection. However, the generalization ability of the deep network model for unknown materials, and the computational complexity of the network, need to be further improved. A new lightweight fingerprint liveness detection network is here proposed to distinguish fake fingerprints from real ones. The method includes mainly foreground extraction, fingerprint image blocking, style transfer based on CycleGan and an improved ResNet with multi-head self-attention mechanism. The proposed method can effectively extract ROI and obtain the end-to-end data structure, which increases the amount of data. For false fingerprints generated from unknown materials, the use of CycleGan network improves the model generalization ability. The introduction of Transformer with MHSA in the improved ResNet improves detection performance and reduces computing overhead. Experiments on the LivDet2011, LivDet2013 and LivDet2015 datasets showed that the proposed method achieves good results. For example, on the LivDet2015 dataset, our methods achieved an average classification error of 1.72 across all sensors, while significantly reducing network parameters, and the overall parameter number was only 0.83 M. At the same time, the experiment on small-area fingerprints yielded an accuracy of 95.27%. Full article

A rich and effective dataset is an important foundation for the development of AI algorithms, and the quantity and quality of the dataset determine the upper limit level of the algorithms. For aerospace remote sensing datasets, due to the high cost of data collection and susceptibility to meteorological and airway conditions, the existing datasets have two problems: firstly, the number of datasets is obviously insufficient, and, secondly, there is large unevenness between different categories in datasets. One of the effective solutions is to use neural networks to generate fake data by learning from real data, but existing methods still find difficulty in generating remote sensing sample images with good texture detail and geometric distortion. To address the shortcomings of existing image generation algorithms, this paper proposes a gradient structure information-guided attention generative adversarial network (SGA-GAN) for remote sensing image generation, which contains two innovative initiatives: on the one hand, a learnable gradient structure information extraction branch network can be added to the generator network to obtain complex structural information in the sample image, thus alleviating the distortion of the sample geometric structure in remote sensing image generation; on the other hand, a multidimensional self-attention feature selection module is proposed to further improve the quality of the generated remote sensing images by connecting cross-attentive modules as well as spatial and channel attention modules in series to guide the generator to better utilize global information. The algorithm proposed in this paper outperformed other methods, such as StyleGAN-XL and FastGAN, in both the qualitative and quantitative evaluation, whereby the FID on the DOTA dataset decreased by 23.927 and the IS was improved by 2.351. The comparison experiments show that the method proposed in this paper can generate more realistic sample images, and images generated by this method can improve object detection metrics by increasing the number of single-category datasets and the number of targets in fewer categories in multi-category datasets, which means it can be effectively used in the field of intelligent processing of remote sensing images. Full article

This research proposes a novel technique for fake news classification using natural language processing (NLP) methods. The proposed technique, TwIdw (Term weight–inverse document weight), is used for feature extraction and is based on TfIdf, with the term frequencies replaced by the depth of the words in documents. The effectiveness of the TwIdw technique is compared to another feature extraction method—basic TfIdf. Classification models were created using the random forest and feedforward neural networks, and within those, three different datasets were used. The feedforward neural network method with the KaiDMML dataset showed an increase in accuracy of up to 3.9%. The random forest method with TwIdw was not as successful as the neural network method and only showed an increase in accuracy with the KaiDMML dataset (1%). The feedforward neural network, on the other hand, showed an increase in accuracy with the TwIdw technique for all datasets. Precision and recall measures also confirmed good results, particularly for the neural network method. The TwIdw technique has the potential to be used in various NLP applications, including fake news classification and other NLP classification problems. Full article