Attentive statement fraud detection: Distinguishing multimodal financial data with fine-grained attention

Wang, Gang; Ma, Jingling; Chen, Gang

doi:10.1016/j.dss.2022.113913

Cited by 9 publications

(3 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ashtiani et al [1] propose a novel methodology leveraging generative adversarial networks (GANs) and ensemble models, adept at addressing the scarcity of non-fraudulent samples while concurrently managing the complexity of high-dimensional feature spaces. Wang et al [3] further adopted multi-modalities data to detect financial statement fraud. However, these typical algorithms often require extensive data preprocessing and may exhibit inefficiencies when dealing with large datasets, leading to less-than-ideal identification results overall.…”

Section: Related Workmentioning

confidence: 99%

WOA-XGBoost-based financial statement fraud detection

Dou,

Liu,

Yin

2024

Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024)

View full text Add to dashboard Cite

Financial statement fraud detection is a crucial topic in maintaining financial security. To optimize parameter tuning and improve model performance, we innovatively combined Whale Optimization Algorithm (WOA) with Extreme Gradient Boosting Algorithms (XGBoost) to construct the WOA-XGBoost model. We first construct a XGBoost based classification model based on XGBoost, and adopt the WOA algorithm to adaptively optimize the XGBoost parameters. Compared to manually adjusting the model training parameters, WOA-XGBoost adaptively searches for the optimal parameters. The experimental results presents a noticeable improvement in F1-Score for WOA-XGBoost. Meanwhile, WOA-XGBoost achieved the highest accuracy among the algorithms participating in the comparative experiment, which demonstrating the superior performance of the proposed method.

show abstract

Section: Related Workmentioning

confidence: 99%

WOA-XGBoost-based financial statement fraud detection

Dou,

Liu,

Yin

2024

Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024)

View full text Add to dashboard Cite

show abstract

“…Indeed, this paper is poised to catalyze advancements in engineering technology. One intuitive application lies in financial fraud detection and data manipulation verification [1,3,4,[6][7][8]]. It's conceivable that we can boldly embark on collaborative efforts leveraging the insights derived from Benford's law, Zipf's law, and Gamma distribution, all residing within the numerical data.…”

Section: Implicationmentioning

confidence: 99%

“…In the contemporary discourse concerning facets encompassing corporate banking credit, investment due diligence, supplier admission audits, and national regulatory governance, the indispensable role is assumed by the research of their financial statements [1][2][3]. Financial statements constitute written documents that encapsulate an enterprise's financial position and operational achievements, meticulously chronicling aspects such as assets and liabilities, profitability, cash flows, and changes in owner's equity.…”

Section: Introductionmentioning

confidence: 99%

Commercialization of state-owned broadcast networks in China: a case study on the operation of Beijing Renmin Radio Station

Yan¹

View full text Add to dashboard Cite

Visual contents, such as illustrations and images, play a big role in product manual understanding. Existing Product Manual Question Answering (PMQA) datasets tend to ignore visual contents and only retain textual parts. In this work, to emphasize the importance of multimodal contents, we propose a Multimodal Product Manual Question Answering (MPMQA) task. For each question, MPMQA requires the model not only to process multimodal contents but also to provide multimodal answers. To support MPMQA, a large-scale dataset PM209 is constructed with human annotations, which contains 209 product manuals from 27 well-known consumer electronic brands. Human annotations include 6 types of semantic regions for manual contents and 22,021 pairs of question and answer. Especially, each answer consists of a textual sentence and related visual regions from manuals. Taking into account the length of product manuals and the fact that a question is always related to a small number of pages, MP-MQA can be naturally split into two subtasks: retrieving most related pages and then generating multimodal answers. We further propose a unified model that can perform these two subtasks all together and achieve comparable performance with multiple task-specific models. The PM209 dataset is available at https://github.com/AIM3-RUC/MPMQA.

show abstract