4 November 2024 in IEEE Access
With the rise of digital payment methods and the growth in financial transactions, the issue of credit card fraud has become increasingly severe. Traditional fraud detection methods are currently facing challenges such as poor model performance, difficulty in obtaining accurate results, and limitations in distributed deployment. These challenges stem from constantly evolving fraud strategies, higher volumes of transactions, and the complexity of the financial environment. This study proposes a credit card fraud detection algorithm based on Structured Data Transformer (SDT) and federated learning, which leverages the advanced capabilities of the Transformer model in deep learning. First, we organize credit card data into sequences and introduce a special, learnable token at the beginning of each sequence for classification purposes. Thanks to the attention mechanism of the Transformer, the model can automatically highlight important features in the data, significantly improving the accuracy of fraud detection. Second, addressing the complex financial environment and concerns about financial data privacy, we introduce a federated learning architecture to deploy the SDT model across different banks in a distributed manner. Momentum updates are used for model parameter updates during training, which enhance model performance and ensure data privacy between banks. Lastly, we conducted experimental validation on two financial datasets of different scales. The results on Dataset 1 and Dataset 2 show that our proposed SDT model surpasses traditional detection methods in terms of AUC-PR values (0.882, 0.816) and AUC-ROC values (0.982, 0.994). By integrating federated learning and deploying and testing the two datasets in a distributed environment, the AUC-PR values (0.884, 0.892) and AUC-ROC values (0.963, 0.998) can be further improved.
11 April 2024 in IEEE Access
Financial fraud cases causing serious damage to the interests of investors are not uncommon. As a result, a wide range of intelligent detection techniques are put forth to support financial institutions’ decision-making. Currently, existing methods have problems such as poor detection accuracy, slow inference speed, and weak generalization ability. Therefore, we suggest a distributed knowledge distillation architecture for financial fraud detection based on Transformer. Firstly, the multi-attention mechanism is used to give weights to the features, followed by feed-forward neural networks to extract high-level features that include relevant information, and finally neural networks are used to categorize financial fraud. Secondly, for the problem of inconsistent financial data indicators and unbalanced data distribution focused on different industries, a distributed knowledge distillation algorithm is proposed. This algorithm combines the detection knowledge of the multi-teacher network and migrates the knowledge to the student network, which detects the financial data of different industries. The final experimental results show that the proposed method outperforms other methods in terms of F1 score (92.87%), accuracy (98.98%), precision (81.48%), recall (95.45%), and AUC score (96.73%) when compared to the traditional detection methods.
20 August 2018 in Nanoscale and Microscale Thermophysical Engineering
Megahertz is the highest femtosecond laser repetition rate that the state-of-the art technology can achieve. In this article, a single femtosecond laser pulse is burst into multiple femtosecond laser pulses to process aluminum. The temporal gap between two consecutive burst pulses is 2 picoseconds, which is much shorter than the temporal gap between two consecutive pulses at the repetition rate of megahertz. By taking the thermophysical scenarios of femtosecond laser induced of electron thermalization, electron heat conduction, electron–phonon-coupled heat transfer and atomic motion into account, a multiscale framework integrating ab initio quantum mechanical calculation, molecular dynamics and two-temperature model are constructed. The effect of femtosecond laser pulse number on the incubation phenomenon is studied. Comparing with the single pulse-processing aluminum film, the femtosecond laser in burst mode leads to smaller thermal stress, which is favorable to reduce the thermal mechanical damage of the material beneath the laser-irradiated surface. Appreciable differences among the simulation results by using electron thermophysical parameters from ab initio quantum mechanical calculation and those from experimental measurement, empirical estimation and calculation are found, indicating the essentials to precisely model the electron thermal response subject to femtosecond laser excitation.
19 September 2019 in Journal of Electronic Packaging
Quantum dot (QD) attracts great attention in light-emitting diode (LED) packaging for high-quality light sources, while it leads to low light efficiency due to the significantly high reabsorption loss between QDs. Accordingly, we experimentally study the reabsorption properties of QD color convertors (QCCs) for LED packaging considering various thicknesses and concentrations under different injection current. The results indicate the QCC configuration with a small thickness and large concentration can have the same absorption ability for chip light as that with the opposite configuration, resulting in the same QD light proportion. However, the QCC configuration having smaller thickness is more useful to decrease the reabsorption loss, leading to higher radiant power (an increase of larger than 37.2%). Moreover, it is essential to gain a high radiant power of QD light with small reabsorption loss, which can be realized by combining QCCs with a low QD content and a source with a large injection current. Based on this simple and effective approach, a conversion loss smaller than 20%, close to their quantum yield, can be achieved, which is approximately four times smaller than that gained by QCCs with a high QD content. However, it introduces additional radiant power of chip light, suppressing further improvement in the QD light proportion. Much work is still required to make full use of the redundant chip light. This study provides a better understanding of the reabsorption properties of QCCs and can significantly accelerate their applications in illumination and display applications.
1 December 2016 in Solid-State Electronics
14 July 2024 in Applied Thermal Engineering