Multiple Transformations Research Articles

Application of Machine Learning Models in Fraud Detection in Financial Transactions

Introduction: fraud detection in financial transactions has become a critical concern in today's financial landscape. Machine learning techniques have become a key tool for fraud detection given their ability to analyze large volumes of data and detect subtle patterns.Objective: evaluate the performance of machine learning techniques such as Random Forest and Convolutional Neural Networks to identify fraudulent transactions in real time.Methods: a real-world data set of financial transactions was obtained from various institutions. Data preprocessing techniques were applied that include multiple imputation and variable transformation. Models such as Random Forest, Convolutional Neural Networks, Naive Bayes and Logistic Regression were trained and optimized. Performance was evaluated using metrics such as F1 score.Results: random Forests and Convolutional Neural Networks achieved an F1 score greater than 95% on average, exceeding the target threshold. Random Forests produced the highest average F1 score of 0,956. It was estimated that the models detected 45 % of fraudulent transactions with low variability.Conclusions: the study demonstrated the effectiveness of machine learning models, especially Random Forests and Convolutional Neural Networks, for accurate real-time fraud detection. Its high performance supports the application of these techniques to strengthen financial security. Future research directions are also discussed

Energy consumption analysis of power grid distribution transformers based on an improved genetic algorithm

With the promotion of energy transformation, the utilization ratio of electrical power is progressively rising. Since electrical power is challenging to store, real-time production and consumption become imperative, imposing significant demands on the dependability and operational efficiency of electrical power apparatus. Suppose the load distribution among multiple transformers within a transformer network exhibits inequality. In such instances, it will amplify the total energy consumption during the voltage conversion process, and local, long-term high-load transformer networks become more susceptible to failures. In this article, we scrutinize the matter of transformer energy utilization in the context of electricity transmission within grid systems. We propose a methodology grounded on genetic algorithms to optimize transformer energy usage by dynamically redistributing loads among diverse transformers based on their operational status monitoring. In our experimentation, we employed three distinct approaches to enhance energy efficiency. The experimental findings evince that this approach facilitates swifter attainment of the optimal power level and diminishes the overall energy consumption during transformer operation. Moreover, it exhibits a heightened responsiveness to fluctuations in power demand from the electrical grid. Experimental results manifest that this technique can truncate monitoring time by 27% and curtail the overall energy consumption of the distribution transformer network by 11.81%. Lastly, we deliberate upon the potential applications of genetic algorithms in the realm of power equipment management and energy optimization issues.

Identifying Winter Wheat Using Landsat Data Based on Deep Learning Algorithms in the North China Plain

The North China Plain (NCP) represents a significant agricultural production region in China, with winter wheat serving as one of its main grain crops. Accurate identification of winter wheat through remote sensing technology holds significant importance in ensuring food security in the NCP. In this study, we have utilized Landsat 8 and Landsat 9 imagery to identify winter wheat in the NCP. Multiple convolutional neural networks (CNNs) and transformer networks, including ResNet, HRNet, MobileNet, Xception, Swin Transformer and SegFormer, are used in order to understand their uncertainties in identifying winter wheat. At the same time, these deep learning (DL) methods are also compared to the traditional random forest (RF) method. The results indicated that SegFormer outperformed all methods, of which the accuracy is 0.9252, the mean intersection over union (mIoU) is 0.8194 and the F1 score (F1) is 0.8459. These DL methods were then applied to monitor the winter wheat planting areas in the NCP from 2013 to 2022, and the results showed a decreasing trend.

Measurement of Molten Steel Level Using a Single Camera in Top Side-Pouring Twin-Roll Casting

The molten steel level in twin-roll strip casting (TRC) has a significant impact on the heat transfer process between the molten steel and the rolls, as well as the subsequent solidification process of the steel. Therefore, ensuring a specific and stable molten steel level is crucial for the quality of as-casting strips. To achieve this, a precise and real-time molten steel level detection system is required. This paper utilizes machine vision technology to measure the molten steel level. A general mathematical model for the molten steel level in the TRC process is established. An image processing method for measuring the molten steel level using a single camera is proposed, including image segmentation, edge detection, and multiple coordinate transformations of the molten pool contour. The adverse effect of the inlet or nozzle is taken into account. Experimental measurements were conducted, and the results indicate that a single camera can accurately measure the molten steel level. Potential sources of error or limitations that may impact the accuracy of the proposed method is discussed.

Multiple feature fusion transformer for modeling penicillin fermentation process with unequal sampling intervals.

The quality prediction of batch processes is an important task in the field of biological fermentation. However, dynamic nonlinearity, unequal sampling intervals, uneven duration, and multiple features of a batch process make this task challenging. Thus, the multiple-feature fusion transformer (MFFT) model is proposed for the time series quality prediction of a batch process. First, the application of sequence-to-sequence architecture enables MFFT to perform a wide range of sequence prediction tasks. Second, the transformer parallel operation model imposes no rigid requirement for the order of sequence input, allowing the model to deal with problems of unequal interval sampling and utilize the sequence information. Third, MFFT integrates a pretrained ResNet50 as a mycelium status classifier for fusing image information into the features. Moreover, a multiple-feature encoding structure is proposed to integrate sampling time and mycelium status. Finally, multiple tasks in penicillin fermentation have shown that MFFT significantly outperforms existing methods for time series prediction.

Unraveling Multiple Pathways of Electron Donation from Phenolic Moieties in Natural Organic Matter.

Natural organic matter (NOM) exhibits a distinctive electron-donating capacity (EDC) that serves a pivotal role in the redox reactions of contaminants and minerals through the transformation of electron-donating phenolic moieties. However, the ambiguity of the molecular transformation pathways (MTPs) that engender the EDC during NOM oxidation remains a significant issue. Here, MTPs that contribute to EDC were investigated by identifying the oxidized products of phenolic model compounds and NOM samples in direct or mediated electrochemical oxidation (DEO or MEO, respectively) using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS). It was found that the oxidation of newly formed phenolic-OH (ArOH) and the oxidative coupling reaction of the phenoxy radical are the main MTPs that directly contribute to EDC, in addition to the transformation of hydroquinones to quinones. Notably, the oxidative coupling reaction of ArOH contributed at least 22-42% to the EDC. Ferulic acid-like structures can also directly contribute to EDC by incorporating H2O into their acrylic substituents. Furthermore, the opening of C rings can indirectly attenuate the EDC through structural alterations in the electron-donating process of NOM. Decarboxylation can either weaken or enhance the EDC depending on the structure of the phenolic moieties in NOM. These findings suggest that the EDC of NOM is a comprehensive result of multiple NOM MTPs, involving not only ArOH oxidation but also the addition of H2O to olefinic bonds and bond-breaking reactions. Our work provides molecular evidence that aids in the comprehension of the multiple EDC-associated transformation pathways of NOM.

A Specific Emitter Identification System Design for Crossing Signal Modes in the Air Traffic Control Radar Beacon System and Wireless Devices.

To improve communication stability, more wireless devices transmit multi-modal signals while operating. The term 'modal' refers to signal waveforms or signal types. This poses challenges to traditional specific emitter identification (SEI) systems, e.g., unknown modal signals require extra open-set mode identification; different modes require different radio frequency fingerprint (RFF) extractors and SEI classifiers; and it is hard to collect and label all signals. To address these issues, we propose an enhanced SEI system consisting of a universal RFF extractor, denoted as multiple synchrosqueezed wavelet transformation of energy unified (MSWTEu), and a new generative adversarial network for feature transferring (FTGAN). MSWTEu extracts uniform RFF features for different modal signals, FTGAN transfers different modal features to a recognized distribution in an unsupervised manner, and a novel training strategy is proposed to achieve emitter identification across multi-modal signals using a single clustering method. To evaluate the system, we built a hybrid dataset, which consists of multi-modal signals transmitted by various emitters, and built a complete civil air traffic control radar beacon system (ATCRBS) dataset for airplanes. The experiments show that our enhanced SEI system can resolve the SEI problems associated with crossing signal modes. It directly achieves 86% accuracy in cross-modal emitter identification using an unsupervised classifier, and simultaneously obtains 99% accuracy in open-set recognition of signal mode.

MIL-ViT: A multiple instance vision transformer for fundus image classification

Despite the great success of deep learning approaches, retinal disease classification is still challenging as the early-stage pathological regions of retinal diseases may be extremely tiny and subtle, which are difficult for networks to detect. The feature representations learnt by deep learning models focusing more on the local view may lead to indiscriminative semantic-level representation. On the contrary, if they focus more on the global semantic-level, they may ignore the discerning subtle local pathological regions. To address this issue, in this paper, we propose a hybrid framework, combining the strong global semantic representation learning capability of the vision Transformer (ViT) and the excellent capacity of local representation extraction from the conventional multiple instance learning (MIL). Particularly, a multiple instance vision Transformer (MIL-ViT) is implemented, where the vanilla ViT branch and the MIL branch generate semantic probability distributions separately, and a bag consistency loss is proposed to minimize the difference between them. Moreover, a calibrated attention mechanism is developed to embed the instance representation into the bag representation in our MIL-ViT. To further improve the feature representation capability for fundus images, we pre-train the vanilla ViT on a large-scale fundus image database. The experimental results validate that the generalization capability of the model using our pre-trained weights for fundus disease diagnosis is better than the one using ImageNet pre-trained weights. Extensive experiments on four publicly available benchmarks demonstrate that our proposed MIL-ViT outperforms latest fundus image classification methods, including various deep learning models and deep MIL methods. All our source code and pre-trained models are publicly available at https://github.com/greentreeys/MIL-VT.

Recent Applications of Multiple Hypergeometric Transformations in Function Spaces and Associated Reduction Formulas

In this research paper we have obtained recent applications of multiple hypergeometric transformations in function spaces and associated reductions formulas of Kampé de Ferlet’s, Exton’s double series, and Srivastava triple series. Further, we have presented some interesting integrals involving the Lauricella’s , Lauricella-Saran’s , etc., Exton’s , Horn’s and Jain’s and generalized hypergeometric series arise as special cases of our main results. Also, some special cases were considered as an application of presented integral functions in function spaces.

Effect of Multiple Reverse Transformation Treatments on Grain Refinement and Mechanical Properties of Biomedical Co-Cr-Mo-N Alloys Fabricated by Electron Beam Melting.

We investigated the improvement of mechanical properties of biograde Co-28Cr-6Mo-0.11N alloy prepared by electron beam melting through grain refinement via multiple reverse transformations. While the effects of single and double reverse transformation treatments on the microstructure have been investigated in previous studies, we investigated the effects of multiple reverse transformation heat treatments. The particle size was refined to 1/4, and the yield strength, tensile silence strength, and elongation were enhanced to 655 MPa, 1234 MPa, and 45%, respectively, satisfying ASTM F75 standards. Moreover, a mixed phase of ε and γ was observed to provide higher yield strength than a single γ structure. The dominant behavior in the γ → ε phase transformation at 1073 K was obvious. Grain growth was suppressed by the grain-boundary pinning effect of the Cr2N phase during reverse transformation to the γ phase. Because no fracture was caused by precipitates such as σ, η, and Cr2N phases, the influence of the precipitates on the tensile properties was small.

Joint time domain nonlinear post-distortion scheme for reconstruction of distorted signals

Reconstruction of distorted signals (RODS), initially proposed in the frequency domain (FD), has been proved to be with high efficiency in depressing the distortion caused by the nonlinear power amplifier (PA). However, multiple high-dimensional transformation matrices are needed to reconstruct the FD distortion matrix, so high complexity is required. In this paper, we propose a new unified time domain (TD) nonlinear post-distortion scheme based on RODS, where a joint distortion matrix containing both PA and multipath channel coefficients is presented. When carrying out the least square inverse operation, the nonlinearity as well as channel fading effect can be eliminated simultaneously. By adjusting the elements in the distortion matrix according to the characteristics of PA and multipath channels, the proposed TD-RODS can also be suitable for flat fading with memoryless or memory nonlinear PAs. Further, the equivalence between the TD- and FD-RODS is presented, and then computer simulations are adopted to verify our analysis. In the proposed TD-RODS, the nonlinear distortion matrix is constructed more straightforward in a more cost-efficiency way, thus the overall computations is effectively reduced.

Multiple templates transformer for visual object tracking

Matching the similarity between a template and search region is crucial in Siamese trackers. However, due to the limited information provided by a fixed template, existing trackers are not robust enough in complex scenarios, such as severe deformation, background clutters, out-of-view, illumination variation, low resolution, scale variation, fast motion, and full occlusion. Therefore, it is essential to use an informative template. Additionally, since the Transformer has superior model capability compared to traditional cross-correlation in tracking, some Siamese trackers have integrated Transformers and achieved exceptional performance. In this paper, we present a novel tracking architecture with Multiple Templates Transformer (MTT) to address the above issues. By utilizing multiple templates, the proposed method can grasp more contextual information and historical changes about the target, which can be leveraged to enhance the response in the search region using an encoder-decoder framework. We also explore different mechanisms to fuse templates effectively to achieve higher accuracy. We evaluate MTT in several famous benchmarks such as GOT-10k, TrackingNet, UAV123, OTB2015, VOT2018, and LaSOT. Extensive experimental results indicate that our tracker is capable of achieving better robustness in the face of different challenges while maintaining a considerable real-time speed.

Obstacles To The Internationalization Of Moroccan SMEs: Case Studies

The environment in which companies are immersed is characterized by multiple transformations that can be described as profound, especially with the successive reduction of tariff and non-tariff barriers, the acceleration of competition, and changes in the business world (Nummela, 2004). Several countries have set up free-trade zones to facilitate integrating companies into the global economy ((Raynard & Forstater, 2002). At this level, the internationalization of companies is seen as a solution for reaping the benefits of globalization, in particular, taking advantage of the exploitation of international value chains (Williamson, 1995) or progressively seeking continuous growth (Dunning, 2000). Indeed, SMEs are also concerned by this logic (Marchesnay et al., 2015); (Olinga & Bikoah, 2022) since internationalization represents for small and medium-sized enterprises a primordial factor for growth, development, and sustainability (Cerrato & Piva, 2010). In Morocco, SMEs play an essential role in the dynamics of the national economy. According to the figures, they account for over 95% of value-added companies, 40% of private investment, and around 30% of Moroccan exports (Bentaleb & Louitri, 2011). Although SMEs are less active internationally, they are very successful when going international (Bongolomba Isoketsu, 2017). They are obliged to choose different approaches to internationalization than large companies (Zacharakis, 1997). That is why choosing an international strategy is essential for them to own a place and a competitive position, but on the condition that they think and pursue a more global management vision (Dutot et al., 2014). Keywords: SMEs, case studies, Internationalization, Obstacles, SMEs

Bathing through Time and Landscape: Everyday Encounters and Reconstructions at Huaqing Hot Springs (1000–1900)

Abstract Despite the long-standing interest in hot springs in China, this subject has remained understudied. This article explores how the most famous hot springs in China, the Huaqing Hot Springs 華清池 at Mount Li 驪山, were selectively reused and negotiated by different social groups from the tenth to the nineteenth century. It retraces the historical process of reconstructions at the Huaqing Hot Springs by local officials, religious practitioners, literati, and local inhabitants after the Huaqing palaces were abandoned by the Tang central court. In the post-Tang life of the Huaqing Hot Springs, the long-term co-existence of activities between various social classes and across the gender divide are deeply rooted in the local everyday usage of the hot springs. This article reveals multiple transformations within continuity in reimaging and reconstructing the hot springs landscape in Chinese history.

MSVT: Multiple Spatiotemporal Views Transformer for DeepFake Video Detection

Recently, DeepFake videos have developed rapidly, causing new security issues in society. Due to the rough spatiotemporal view, existing video-based detection methods struggle to capture fine-grained spatiotemporal information, resulting in limited generalization ability. In addition, although the transformer has achieved great success in the past few years, the application of transformer on deepfake video detection still needs to be studied. To solve this problem, in this paper, we propose a novel Multiple Spatiotemporal Views Transformer (MSVT) with Local Spatiotemporal View (LSV) and Global Spatiotemporal View (GSV), to mine more detailed spatiotemporal information. Firstly, for establishing the LSV, different from existing works that sparsely sample a single frame to build the input sequence, we employ the local-consecutive temporal view to capture vital dynamic inconsistency. Furthermore, the extracted frame features within each group are fed to the temporal transformer followed by the feature fusion module, to generate group-level spatiotemporal features. Then, we further establish Global Spatiotemporal View (GSV) by feeding all the frame features within the whole video to the temporal transformer followed by the feature fusion module. Finally, we propose a novel global-local transformer (GLT) to effectively integrate these multi-level features for mining more subtle and comprehensive features. Extensive experiments on six large datasets demonstrate that our MSVT outperforms state-of-the-art detection methods.

Dynamic Output Feedback Fault Tolerant Control for Switched Fuzzy Systems With Fast Time Varying and Unbounded Faults

This paper focuses on the fault reconstruction and fault-tolerant control problems of switched fuzzy systems with actuator and sensor faults, where the faults may be fast time-varying and unbounded. The multiple coordinate transformations are introduced, then the reduced-order observer based fault reconstruction method is considered, which can reconstruct the system state, actuator and sensor faults, simultaneously. For the case that there is no disturbance, the LMI conditions are achieved to insure that for any switching signal, the error dynamics can be exponentially stable. Utilizing the estimated fault information, the fuzzy dynamic output feedback fault tolerant controllers are designed to compensate the fault effect and stabilize the systems. The root-mean-square gain is considered to reduce the disturbance influence on fault estimation and fault-tolerant control, in which the zero initial condition is not needed. Finally, simulations are conducted to demonstrate the effectiveness of the considered techniques.

COVID-19 Vaccine Adverse Event Detection Based on Multi-Label Classification With Various Label Selection Strategies.

Analyzing massive VAERS reports without medical context may lead to incorrect conclusions about vaccine adverse events (VAE). Facilitating VAE detection promotes continual safety improvement for new vaccines. This study proposes a multi-label classification method with various term-and topic-based label selection strategies to improve the accuracy and efficiency of VAE detection. Topic modeling methods are first used to generate rule-based label dependencies from Medical Dictionary for Regulatory Activities terms in VAE reports with two hyper-parameters. Multiple label selection strategies, namely one-vs-rest (OvsR), problem transformation (PT), algorithm adaption (AA), and deep learning (DL) methods, are used in multi-label classification to examine the model performance, respectively. Experimental results indicated that the topic-based PT methods improve the accuracy by up to 33.69% using a COVID-19 VAE reporting data set, which improves the robustness and interpretability of our models. In addition, the topic-based OvsR methods achieve an optimal accuracy of up to 98.88%. The accuracy of the AA methods with topic-based labels increased by up to 87.36%. By contrast, the state-of-art LSTM- and BERT-based DL methods have relatively poor performance with accuracy rates of 71.89% and 64.63%, respectively. Our findings reveal that the proposed method effectively improves the model accuracy and strengthens VAE interpretability by using different label selection strategies and domain knowledge in multi-label classification for VAE detection.

Adopting multiple vision transformer layers for fine-grained image representation

Accurate discriminative regions proposal has an important effect for fine-grained image recognition. The vision transformer (ViT) brings about a striking effect in computer vision duo to its innate muti-head self-attention mechanism. However, the attention maps are gradually similar after certain layers and since ViT adds classification token for perform classification, it is unable to effectively select discriminative image patches for fine-grained image classification. To accurately detect discriminative regions, we propose a novel network AMTrans, which efficiently increases layers to learn diverse features and utilizes integrated raw attention maps to capture more salient feature. Specifically, we employ DeepViT as backbone to solve the attention collapse issue. Then, we fuse each head attention weight within each layer to produce attention weight map. After that, we alternatively use recurrent residual refinement blocks to promote salient feature detection and then utilize semantic grouping method to propose the discriminative feature region. A lot of experiments prove that AMTrans acquires the SOTA performance on three widely used fine-grained datasets under the same settings, involving Stanford-Cars, Stanford-Dogs and CUB-200-2011.

Simultaneous Multispot Temperature Prediction of Traction Transformer in Urban Rail Transit Using Long Short-Term Memory Networks

Hot-spot temperature (HST) is a typical indicator of transformer’s health status, accurate prediction of HST is critical for prognosis and health management (PHM) of traction transformer in urban rail transit (URT), where dramatically fluctuating loads and complex climates pose serious challenges. This paper proposed a novel deep-learning-enabled method to predict multi-spot temperatures of transformer simultaneous using long short-term memory (LSTM) neural networks. Real-world operation data collected from Qingdao Metro over a year-long period were used to train the prediction model and verify its validity. The most appropriate transformer-related parameters were selected as the inputs of model by Pearson correlation coefficients (PCC) to improve the accuracy and efficiency of the model. Besides, the dropout and early stopping techniques are used to prevent model from overfitting. Furthermore, the robustness and versatility of proposed method were verified by testing the data on different seasons of multiple transformers, and the superiority of the method was also proved by comparing with other methods. The results show that the proposed model can achieve accurate on-line temperature prediction of transformer, and the mean-relative-error of sixteen minutes ahead HST prediction is less than 0.26%. The proposed method can provide a reference for PHM of transformers.

Proposal and Assessment of a Multiple Cycle-Continuous Cooling Transformation (MC-CCT) Diagram for Wire Arc Additive Manufacturing of Thin Walls

Continuous cooling transformation (CCT) diagrams of base metals are common in welding. They can be built using physical or numerical simulations, each with advantages and limitations. However, those are not usual for weld metal, considering its variable composition due to the dilution of the weld into the base metal. Wire Arc Additive Manufacturing (WAAM) is a distinctive case in which the interest in materials comparable with weld composition raises attention to estimating their mechanical properties. Notwithstanding, this concept is still not used in WAAM. Therefore, the aim of this work was to address a methodology to raise MC-CCT (Multiple Cycle Continuous Cooling Transformation) diagrams for WAAM by combining physical and numerical simulations. A high-strength low-alloy steel (HSLA) feedstock (a combination of a wire and a shielding gas) was used as a case study. To keep CCT as representative as possible, the typical multiple thermal cycles for additive manufacturing thin walls were determined and replicated in physical simulations (Gleeble dilatometry). The start and end transformations were determined by the differential linear variation approach for each thermal cycle. Microstructure analyses and hardness were used to characterise the product after the multiple cycles. The same CCT diagram was raised by a commercial numerical simulation package to determine the shape of the transformation curves. A range of austenitic grain sizes was scanned for the curve position matching the experimental results. Combining the experimental data and numerically simulated curves made estimating the final CCT diagram possible.