Fraudulent Samples Research Articles

Authentication of edible herbal materials and food products using mass spectrometry based metabolites and inorganic constituents

Medicinal food homologous (MFH) substances not only provide nutrition but also serve as a traditional means to overcome many health issues. Authentication of these products verifies their efficacity and assures consumers of a genuine product. In this review paper, we focus the determination of MFH authenticity including geographical identification and adulteration detection using mass spectrometry (liquid and gas chromatography) based metabolites and inorganic constituents (muti-elements and stable isotopes). The application of these techniques to determine product identification characteristics combined with chemometrics are discussed, along with the limitations of these techniques. Multi-elements, stable isotopes, and metabolite analysis are shown to provide an effective combination of techniques to resolve the origin of various MFH products. Most organic compounds from MFH products are identified using chromatographic separation techniques (HPLC, GC) combined with different detection methods. Chemometric analysis of organic and inorganic fingerprints offers a robust method to detect and classify mislabeled and suspected fraudulent samples of different MFH products.

Ensuring authenticity of cinnamon powder: Detection of adulteration with coffee husk and corn meal using NIR, MIR spectroscopy and chemometrics

Cinnamon (Cinnamomum sp.) is a commonly used spice that is at risk of being counterfeited due to its intricate production process, which can compromise its integrity. This study aimed to develop a rapid detection method using Near Infrared (NIR) and Mid Infrared (MIR) spectroscopy, along with chemometrics, to identify fraud in genuine cinnamon powder. Various techniques, including exploratory, non-targeted, and targeted methods, were used to detect the presence of adulterants such as coffee husks and corn meal in cinnamon, with concentrations ranging from 10% to 50%. The exploratory and non-targeted models successfully identified the potential occurrence of fraud as a result of the addition of coffee husks and corn meal. The models' effectiveness was tested on commercial cinnamon samples, exposing a high proportion of potentially fraudulent samples. Adulteration percentages exceeded 60% for the presence of coffee husks and 50% for corn meal. The results show that NIR and MIR spectroscopy, when used with chemometrics, are effective for monitoring the quality of cinnamon powder.

Signature Fraud Detection Using Deep Learning

The use of signatures for personal identification and verification is quite common. Signatures are validated for many documents such as Bank cheques and legal transactions. The necessity for effective automated solutions for signature verification has grown as signatures are now a prerequisite for both authorization and authentication in legal activities. Two images—the original signature and the test signature—are used as input in this process. To determine whether the signature is fake or not, the characteristics that were extracted are compared, and the difference in error values between them is examined. The growing digital landscape has increased the need for robust and efficient fraud detection systems. This project presents a unique approach to detecting signature fraud using deep learning techniques, specifically employing Siamese Neural Networks, implemented in Python. With a dataset comprising 2149 signature images, encompassing both genuine and fraudulent samples from Dutch users, our model demonstrates remarkable accuracy. The Siamese Neural Network architecture excels in signature verification tasks by learning to distinguish between genuine and fraudulent signatures through contrastive learning. The key achievement of this project is the exceptional accuracy levels attained during the training and validation phases. The model's training accuracy stands at an impressive 98.00%, while the validation accuracy reaches an astonishing 99.00%. These high accuracy rates are a testament to the effectiveness of Siamese Neural Networks in signature fraud detection. In a world where the security of digital signatures is paramount, this project showcases the power of deep learning and Siamese Neural Networks in safeguarding against fraudulent activities. The model's success in accurately distinguishing between authentic and forged signatures offers promising potential for enhancing security measures in various domains, including finance, legal, and document management.

Real-Time Online Banking Fraud Detection Model by Unsupervised Learning Fusion

Digital trades and payments are becoming increasingly popular, as they typically entail monetary transactions. This not only makes electronic transactions more convenient for the end customer, but it also raises the likelihood of fraud. An adequate fraud detection system with a cutting-edge model is critical to minimizing fraud costs. Identifying fraud at the ideal time entails establishing and setting up ubiquitous systems to consume and analyze massive amounts of streaming data. Recent advances in data analytics methods and introducing open-source technology for big data storage and processing opened new options for detecting fraud. This study aims to tackle this critical issue by providing a newly real-time e-transaction fraud detection schema that consolidates the advantages of both unsupervised learners, including autoencoder and extended isolation forests, with cutting-edge big data gadgets such as Spark streaming and sparkling water. It addresses the shortage of non-fraudulent instances and handles the excessive dimension of the set of features. On two real-world transactional datasets, we assess our suggested technique. Compared with other current fraud identification systems, our methodology delivers an elevated accuracy yield of 99%. Furthermore, it outperforms state-of-the-art approaches in reliably identifying fraudulent samples. Doi: 10.28991/HIJ-2024-05-01-014 Full Text: PDF

Predicting Fraud in U.S.-Listed Chinese Companies: An Empirical Analysis based on M-Score and F-Score Models

M-Score and F-Score are financial fraud prediction models that are commonly used in Western capital markets. In this paper, we calculate M-Scores and F-Scores for each dataset of 15 fraudulent Chinese companies (including Hong Kong companies) listed in the U.S. from 2006 to 2020 and compare their M-Scores and F-Scores with 15 companies in the same industry that are not fraudulent. Firstly, we used Student's t-test on data from both models respectively. The result shows that both of them are able to predict fraud well for Chinese stocks listed in the U.S. With the result, we further compare and test the differences between the financial fraudulent samples and the comparison samples in the dimensions of each variable in the M-Score and F-Score models. Finally, we draw a result that they are effective in predicting whether a Chinese company committed financial fraud.

A fast and non-invasive imaging procedure to fight red tuna fraud

Recently, a food fraud regarding tuna samples has been detected in Europe. Some batches of tuna treated with illegal adulterants like vegetable extracts, including beetroot, were detected after several cases of food poisoning. This paper proposes a new fast and non-destructive technique based on smartphone imaging which contributes to the fight against tuna fraud. The color of different samples of three species of tuna (Thunnus thynnus, Thunnus albacares and Thunnus obesus) has been analyzed by a colorimetrically characterized smartphone. Descriptive analysis of the samples shows that different species of tuna can be grouped by their color in CIELAB space, and that samples adulterated with beetroot extracts can be detected by the proposed image processing procedure. A logistic regression model was implemented to distinguish between real and fraudulent samples, and its suitability checked by k-folding cross validation (n = 189), achieving a 0.59% false positive rate with a 10% false negative rate.

Learning from Highly Imbalanced Big Data with Label Noise

This study explores the effects of class label noise on detecting fraud within three highly imbalanced healthcare fraud data sets containing millions of claims and minority class sizes as small as 0.1%. For each data set, 29 noise distributions are simulated by varying the level of class noise and the distribution of noise between the fraudulent and non-fraudulent classes. Four popular machine learning algorithms are evaluated on each noise distribution using six rounds of five-fold cross-validation. Performance is measured using the area under the precision-recall curve (AUPRC), true positive rate (TPR), and true negative rate (TNR) in order to understand the effect of the noise level, noise distribution, and their interactions. AUPRC results show that negative class noise, i.e. fraudulent samples incorrectly labeled as non-fraudulent, is the most detrimental to model performance. TPR and TNR results show that there are significant trade-offs in class-wise performance as noise transitions between the positive and the negative class. Finally, results reveal how overfitting negatively impacts the classification performance of some learners, and how simple regularization can be used to combat this overfitting and improve classification performance across all noise distributions.

Straightforward monitoring of honey with foreign diastase by leveraging the differentiation in LC-UV proteome profiles of authentic and fraudulent samples

Honey adulteration using sugar syrups is ubiquitous and adulterated honey may also be prepared by direct addition of α-amylases from different sources to match the legislation for honey trade. Colorimetric diastase assays may report false negatives in terms of foreign diastase (FD) detection owing to using substrates being prone to be cleaved by any α-amylases. The main objective of this study was to develop a reliable analytical method to determine FD contained honey. For this purpose, intrinsic honey proteins were first isolated, enriched, and cleaned up via experimentally designed serial ultrafiltration methodology. Next, diastase and invertase as abundant honey enzymes were purified simultaneously from obtained crude protein isolate by means of FPLC and novel purification protocol. Total protein and enzyme activity assays along with SDS-PAGE analysis were conducted for purity check. Subsequently, two purity-confirmed enzymes were used as analytical references in method development. With the hyphenation of UV detection, optimized chromatographic resolution, and practical dilute & shoot sample pretreatment, Foreign Amylase Monitoring (FAM) method has been introduced for routine analysis. Purified authentic enzymes were labeled in LC-UV chromatogram of honey proteome profile to distinguish any unnatural enzyme/protein signal. According to the validation of the FAM method, precise (RSDR; 1.27%), and linear (mean R2 >0.995) results were able to obtain. Commercial honey samples (n = 202) collected over 3 consecutive years were probed using the FAM method and industrial α-amylases were also analyzed for verification. The diastase activities were also measured prior to FAM application to elucidate if any positive correlation. The developed method rapidly and reliably detected the presence of FD qualitatively. A total of 74 samples were found to contain FD. It was observed that the presence of FD was correlated in samples with abnormally high diastase activities and an apparent FD peak was identified readily in all enzyme adulterated samples. The developed FAM method and performed trend analysis allowed us to decipher a dramatic increase in FD existence and this revealed the recently emerging problem of honey authenticity.

A Paper-Based Biomimetic Sensing Device for the Discrimination of Original and Fraudulent Cigarette Brands Using Mixtures of MoS2 Quantum Dots and Organic Dyes.

In this study, we investigated the combined effects of MoS2 QDs' catalytic properties and the colorimetric responses of organic reagents to create a sniffing device based on the sensor array concept of the mammalian olfactory system. The aim was to differentiate the volatile organic compounds (VOCs) present in cigarette smoke. The designed optical nose device was utilized for the classification of various cigarette VOCs. Unsupervised Principal Component Analysis (PCA) and supervised Linear Discriminant Analysis (LDA) methods were employed for data analysis. The LDA analysis showed promising results, with 100% accuracy in both training and cross-validation. To validate the sensor's performance, we assessed its ability to discriminate between five cigarette brands, achieving 100% accuracy in the training set and 82% in the cross-validation set. Additionally, we focused on studying four popular Iranian cigarette brands (Bahman Kootah, Omega, Montana Gold, and Williams), including fraudulent samples. Impressively, the developed sensor array achieved a perfect 100% accuracy in distinguishing these brands and detecting fraud. We further analyzed a total of 126 cigarette samples, including both original and fraudulent ones, using LDA with a matrix size of (126 × 27). The resulting LDA model demonstrated an accuracy of 98%. Our proposed analytical procedure is characterized by its efficiency, affordability, user-friendliness, and reliability. The selectivity exhibited by the developed sensor array positions it as a valuable tool for differentiating between original and counterfeit cigarettes, thus aiding in border control efforts worldwide.

Simple Data Augmentation Tricks for Boosting Performance on Electricity Theft Detection Tasks

In practical engineering, electricity theft detection is usually performed on highly imbalanced datasets (i.e., the number of fraudulent samples is much smaller than the benign ones), which limits the accuracy of the classifier. To alleviate the data imbalance problem, this paper proposes simple data augmentation tricks (SDAT) to boost performance on electricity theft detection tasks. SDAT includes five simple but powerful operations: adding noises to electricity consumption readings, drifting values of electricity consumption readings, quantizing electricity consumption readings to a level set, adding a fixed value to electricity consumption readings, and adding changeable values to electricity consumption readings. In addition, eight potential tricks are also mentioned. Numerical simulations are conducted on a real-world dataset. The simulation results show that SDAT can significantly boost the performance of different classifiers, especially for small datasets. Besides, specific suggestions on how to select parameters of SDAT are provided for its migration use to other datasets.

Physicochemical and Biochemical Quality of Raw Milk Sold in Korhogo Town (North of Côte d’Ivoire)

consumed by the local population. To better understand this situation, a study was conducted on the physicochemical and biochemical quality of raw milk marketed in Korhogo town. Thus, 27 samples of raw milk were collected from sellers in the town of Korhogo. Physicochemical and biochemical analyzes of these samples were carried out. In addition, fraudulent samples were counted by determining the rate of wetting and/or skimming. The raw milk produced locally was of good quality since, for the majority of the samples analyzed, the values of the physicochemical and biochemical parameters were within the range of those proposed by the FAO for raw milk. In addition, a minority of sellers (37.04%) of raw milk practiced fraud, and these did so at low rates (< 15%). Ultimately, this raw milk is recommended for consumption from a nutritional point of view. However, this study should be deepened by also evaluating its microbiological quality.

Credit Card Anti-Fraud Prediction Based on Ensemble earning

In recent years, due to the development of Internet technology in the financial industry, cardless and cashless payments have become increasingly popular. At present, people only need to bind their cards to their cell phones to scan the code for payment. In the meanwhile, as the use of credit cards is vigorously promoted nationwide, younger generations relied more on the credit card debt for consumption. However, although these developments have brought a lot of convenience to people's travel and their life, the virtual nature of the Internet and the "Enjoy First, Pay Later" feature of credit cards have led to a significant increase in potential credit risk in the entire transaction, triggering more and more new credit card fraud incidents. While the overall financial environment in China is relatively healthy, the fact that fraudulent behaviors such as credit card overdrafts, counterfeit cards, and credit card frauds are all personal conduct, making it difficult to detect and prevent these individuals from acting illegally in a timely manner. Although fraud represents a small percentage of the overall transaction size, its bad debt losses to merchants and banks can be significant. In order to make a timely prediction of fraud, there are already many single-model machine learning methods such as decision trees and logistic regression. However, the generalization ability of these models is not good enough in the face of complex user behavior features. In addition, since the probability of fraudulent behavior is very small, resulting in few samples that can be trained, even if a model has a high accuracy rate, there is no guarantee that it can accurately predict fraudulent behavior. Therefore, this paper proposes the Adaboost method based on ensemble learning and uses SMOTE oversampling based on a few classes of fraudulent samples of user features to solve the above existing problems.

A novel approach for lavender essential oil authentication and quality assessment

Currently, the quality of lavender (Lavandula angustifolia Mill.) essential oil (LEO) is defined and regulated based on standards and methods established by regulatory authorities. Unfortunately, these existing standards and methods are not sufficient for LEO quality evaluation due to the complexity of LEO and adulteration encouraged by a burgeoning market. This study provides an efficient and reliable method for LEO quality assessment and adulteration detection. After a comprehensive investigation, involving a large set of LEO samples (n = 72) analyzed by multiple techniques (GC/MS, GC/Q-ToF, NMR, and chemometric analysis), a new approach named Q-Index was proposed. Fourteen marker compounds, along with trans-furano-linalool oxide acetate (an indicator of synthetic compound adulteration in LEO), were identified. These marker compounds played significant roles in discriminating the adulterated samples from the authentic LEOs. Calculation of the Q-Index value using the identified marker compounds permitted the detection of fraudulent samples. As demonstrated, all the authentic LEOs exhibited high Q-Index values (>100), whereas the adulterated or poor-quality samples displayed low Q-Index values (<100). The NMR-based chemometric analysis, which served as an independent and complementary approach to the GC/MS and Q-Index methods, was applied in order to assess the validity of the Q-Index method. Overall, the results obtained from different methods were in good agreement. Moreover, compared to the NMR method, the Q-Index approach possessed greater sensitivity in detecting LEO adulteration associated with the addition of synthetic compounds. Results of this study demonstrated that the Q-Index method could be successfully applied for LEO quality assessment and adulteration detection. This approach may have a significant potential to improve quality control for the LEO industry.

Mycobiome analysis for distinguishing the geographical origins of sesame seeds

Sesame (Sesamum indicum) is one of the most widely cultivated crops in Asia and Africa. The identification of the geographical origins of sesame seeds is important for the detection of fraudulent samples. This study was conducted to build a prediction model and suggest potential biomarkers for distinguishing the geographical origins of sesame seeds using mycobiome (fungal microbiome) analysis coupled with multivariate statistical analysis. Sesame seeds were collected from 25 cities in Korea, six cities in China, and five sites in other countries (Ethiopia, India, Nigeria, and Pakistan). According to the expression of fungal internal transcribed spacer (ITS) sequences in sesame seeds, 21 fungal genera were identified in sesame seeds from various countries. The optimal partial least squares-discriminant analysis model was established by applying two components with unit variance scaling. Based on seven-fold cross validation, the predictive model had 94.4% (Korea vs. China/other countries), 91.7% (China vs. Korea/other countries), and 88.9% (other countries vs. Korea/China) accuracy in determining the geographical origins of sesame seeds. Alternaria, Aspergillus, and Macrophomina were suggested as the potential fungal genera to differentiate the geographical origins of sesame seeds. This study demonstrated that mycobiome analysis could be used as a complementary method for distinguishing the geographical origins of raw sesame seeds.

Nontechnical Losses Detection Through Coordinated BiWGAN and SVDD

Nontechnical losses (NTLs) are estimated to be considerable and increasing every year. Recently, high-resolution measurements from globally laid smart meters have brought deeper insights on users' consumption patterns that can be exploited potentially by NTL detection. However, consumption-pattern-based NTL detection is now facing two major challenges: the inefficiency of harnessing high dimensionality and the severe lack of fraudulent samples. To overcome them, an NTL detection model based on deep learning and anomaly detection is proposed in this article, namely bidirectional Wasserstein GAN and support vector data description-based NTL detector (BSBND). Motivated by the powerful ability of generative adversarial networks (GANs) to learn deep representation from high-dimensional distributions of data, in the BSBND, we utilized a BiWGAN for feature extraction from high-dimensional raw consumption records, and a one-class classifier trained only on benign samples-SVDD-is adopted to map features into judgments. Moreover, a novel alternate coordinating algorithm is proposed to optimize the cooperation between the upstream BiWGAN and the downstream SVDD, and also, an interpreting algorithm is proposed to visualize the basis of each fraudulent judgment. Case studies have demonstrated the superiority of the BSBND over the state of the arts, the powerful feature extraction ability of BiWGAN, and also the effectiveness of the proposed coordinating and interpreting algorithms.

A Novel GC-FID Analytical Strategy Reveals Adulteration of Fuels by Vegetable Oil.

Adulteration of diesel fuel by the addition of vegetable oil is a problem that touches several countries around the world and bypasses the complexity of the specifications regarding the automotive diesel fuel distinguishing between customs, fiscal, and commercial-environmental specifications. At an international level, the adoption of the same analysis methods is important for the harmonization processes and the fluidity of the market. In this context, we assist to the diffusion of the same fraud touching several countries or continents since the limit of the same methods are common to many specifications. For several European countries, the revenue lost as a result of this adulteration consists of billions of euros per year. This enormous amount of illicit money feeds organized criminal networks with huge social and environmental damages. This work presents a GC method to quantify vegetable oils in the range of 0.2-7% (w/w) in adulterated diesel fuel, intended for use as motor fuel, with or without extraneous heavy mineral oil. Our study was realized on 100 fraudulent samples collected by the Italian fiscal police in regular oil stations and by the Italian fiscal police and customs officers as a consequence of controls on trucks transporting, in suspicious circumstances, "oil" often declared "antistick and anticorrosion oil" or "protective oil." Conclusions and Highlights: High reliability of results, external validity, ease of replication, simple instrumentation, and sample preparation make this method well suited for a new "smart" protocol for diesel fuels analysis for customs, fiscal, and antifraud purposes.

Corporate Governance Attributes in Fraud Detterence

The failures of corporations such as Enron, WorldCom and HIH Insurance, to name but a few, have heightened investor awareness of the need to not only evaluate company performance, but also to consider the possibility that financial statements may not be a true reflection of company results, as fraudulent activities may have occurred during the reporting period. Since parties who are outside of the firm do not have access to pertinent information, they have to rely upon published financial and non-financial data to form an opinion regarding performance and/or the risk that fraudulent activities may have occurred. The prior literature shows a relationship between weak corporate governance and fraudulent activities, although most if not all of this research relates to Western economies. The differences in institutional setting e.g. cultural values and legal environment in Malaysia would not give the same findings with the study in western economies. Composing of many ethnicities, Malaysia is a multicultural country. With each ethnic group upholding its own culture, values and belief, businesses are conducted according to each ethnic’s culture. The results of this study could shed some light on the influence of institutional setting regarding corporate governance. Companies that were charged with accounting and auditing offences from year 2003 to 2007 were selected as the fraudulent samples. Data was collected from the years these companies were charged with fraud and the year prior to that. Logistic regression analysis was carried out to determine the significant differences between fraudulent and non-fraudulent companies with respect to corporate governance characteristics. The results indicated that the size of the board and the percentage of institutional shareholdings had significant relationships with the likelihood of corporate fraud occurrences consistently across the two-year period studied. The results of this study will assist public, corporate and accounting policy makers in formulating more effective corporate governance mechanisms.

Topological pattern discovery and feature extraction for fraudulent financial reporting

Fraudulent financial reporting (FFR) involves conscious efforts to mislead others regarding the financial condition of a business. It usually consists of deliberate actions to deceive regulators, investors or the general public that also hinder systematic approaches from effective detection. The challenge comes from distinguishing dichotomous samples that have their major attributes falling in the same distribution. This study pioneers a novel dual GHSOM (Growing Hierarchical Self-Organizing Map) approach to discover the topological patterns of FFR, achieving effective FFR detection and feature extraction. Specifically, the proposed approach uses fraudulent samples and non-fraudulent samples to train a pair of dual GHSOMs under the same training parameters and examines the hypotheses for counterpart relationships among their subgroups taking advantage of unsupervised learning nature and growing hierarchical structures from GHSOMs. This study further presents (1) an effective classification rule to detect FFR based on the topological patterns and (2) an expert-competitive feature extraction mechanism to capture the salient characteristics of fraud behaviors. The experimental results against 762 annual financial statements from 144 public-traded companies in Taiwan (out of which 72 are fraudulent and 72 are non-fraudulent) reveal that the topological pattern of FFR follows the non-fraud-central spatial relationship, as well as shows the promise of using the topological patterns for FFR detection and feature extraction.

Discernment of bee pollen loads using computer vision and one-class classification techniques

In this paper, we propose a system for authenticating local bee pollen against fraudulent samples using image processing and classification techniques. Our system is based on the colour properties of bee pollen loads and the use of one-class classifiers to reject unknown pollen samples. The latter classification techniques allow us to tackle the major difficulty of the problem, the existence of many possible fraudulent pollen types.Also presented is a multi-classifier model with an ambiguity discovery process to fuse the output of the one-class classifiers. The method is validated by authenticating Spanish bee pollen types, the overall accuracy of the final system of being 94%. Therefore, the system is able to rapidly reject the non-local pollen samples with inexpensive hardware and without the need to send the product to the laboratory.