
==== Front
PeerJ Comput Sci
PeerJ Comput Sci
peerj-cs
PeerJ Computer Science
2376-5992
PeerJ Inc. San Diego, USA

cs-2281
10.7717/peerj-cs.2281
Algorithms and Analysis of Algorithms
Artificial Intelligence
Data Mining and Machine Learning
Scientific Computing and Simulation
Social Computing
Intelligent accounting optimization method based on meta-heuristic algorithm and CNN
Dong Yanrui Yanrui19937896822@126.com

School of Accounting, Zhengzhou Vocational College of Finance and Taxation, Zhengzhou, Henan, China
Aleem Muhammad
20 9 2024
2024
10 e228130 5 2024
1 8 2024
©2024 Dong
2024
Dong
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ Computer Science) and either DOI or URL of the article must be cited.

The evolution of social intelligence has led to the adoption of intelligent accounting practices in enterprises. To enhance the efficiency of enterprise accounting operations and improve the capabilities of accountants, we propose an intelligent accounting optimization approach that integrates meta-heuristic algorithms with convolutional neural networks (CNN). First, we enhance the CNN framework by incorporating document and voucher information into accounting audits, creating a multi-modal feature extraction mechanism. Utilizing these multi-modal accounting features, we then introduce a method for assessing accounting quality, which objectively evaluates financial performance. Finally, we propose an optimization technique based on meta-heuristic principles, combining genetic algorithms with annealing models to improve the accounting system. Experimental results validate our approach, demonstrating an accuracy of 0.943 and a mean average precision (mAP) score of 0.812. This method provides technological support for refining accounting audit mechanisms.

Accounting optimization
Meta-heuristic algorithm
CNN
The author received no funding for this work.
==== Body
pmcIntroduction

In the economy’s dynamic evolution, the complexity of businesses continues to grow. Globalization exposes enterprises to expanded cross-border opportunities, necessitating adaptation to diverse regional accounting standards. Concurrently, evolving regulations impose new demands on financial reporting. Advances in information technology further enhance the efficiency of financial data processing and reporting. In this context, studying and continually optimizing accounting methods becomes crucial for financial transactions to accommodate larger and more intricate business environments. This effort contributes to the formulation of authoritative financial standards and enhances the comparability and transparency of corporate financial reporting (Wang et al., 2022a; Zhou, 2022).

Accounting constitutes an essential facet of business operations. Through deep exploration of accounting methods, more precise and credible financial reporting practices can be identified, bolstering the accuracy and credibility of financial data. Robust accounting practices facilitate the provision of comprehensive and transparent financial information, thereby serving as a foundation for intelligent decision-making by enterprise management and investors, thereby promoting efficient economic operations (Hu, 2022). Moreover, a thorough examination of accounting methods aids in establishing consistent international financial standards, fostering global financial information coordination, and enhancing global economic system stability. Amidst an ever-evolving economic landscape, an in-depth study of accounting methods enables enterprises to adapt and enhance their competitiveness effectively (Chen & Chao, 2022; Cui & Verma, 2022). Overall, the study of accounting methods enhances financial data precision within enterprises and contributes to the broader development of the financial sector within economic and regulatory frameworks.

Studying accounting methods involves numerous difficulties and challenges, which increase the complexity researchers face when proposing new methods or improving existing ones (Berdiyeva, Islam & Saeedi, 2021). Firstly, enterprises’ business environment is increasingly complex, with diversified business models and transaction structures. Accounting methods must, therefore, adapt flexibly to this diversity and ensure that research outcomes are applicable across different industries and types of enterprises (Mancini, Lombardi & Tavana, 2021; Zhuo, 2022; Yuan, 2023). Secondly, regulations and accounting standards in various countries and regions are continually evolving. Researchers must stay updated with these changes to ensure that proposed accounting methods align with the latest regulatory requirements and maintain compliance. Lastly, in the era of big data, the vast and diverse datasets generated by enterprises pose a significant challenge in processing and analyzing data to extract useful financial information (Wu, 2021; Li Cui, 2022).

To overcome these challenges, researchers integrate expertise in accounting and adeptly apply advanced machine learning techniques. For instance, Hernes & Sobieska-Karpińska (2016) developed a consensus-based group accounting support system evaluated using data from the Warsaw Stock Exchange. Qu (2023) structured an intelligent accounting system into acquisition, storage, and analysis layers, enhancing decision-making capabilities through intelligent accounting evaluation systems. Wang, Zhang & Tao (2021) proposed an intelligent financial sharing platform based on convolutional neural network (CNN) to improve accounting efficiency, minimize costs, and enhance compliance. Chen & Metawa (2020) introduced a cloud-based enterprise financial management system that processes accounting data. Desyatnyuk, Muravskyi & Shevchuk (2021) innovated with unmanned and expert accounting models based on artificial intelligence (AI) theories. Alazzazbi, Mustafa & Karage (2023) constructed a financial system integrating accounting and risk management, while Guo (2021) utilized data mining to design a comprehensive financial analysis system. Lai (2022) explored an intelligent financial system framework and introduced digital mining-based accounting support. Ding (2022) developed an intelligent financial management model using digital technology. These innovative approaches leverage advanced technologies to address the complexities and demands of modern accounting, aiming to enhance financial management practices and support decision-making in enterprise contexts.

However, with the advancement of the times and the deepening of globalization, the business environment is becoming increasingly complex. In this rapidly changing era, the increasingly diversified business models and transaction structures of enterprises have made traditional accounting methods accurately capture and reflect the true financial status of enterprises. In a rapidly changing business environment, enterprises need not just simple financial data recording but also a financial system that can reflect the operational status of the enterprise in a timely and accurate manner and provide strong support for decision-making. This article proposes an innovative intelligent accounting optimization method to address this challenge. This method is based on the powerful capabilities of meta-heuristic algorithms and CNN. Meta-heuristic algorithms possess the characteristics of global search and optimization, enabling them to find the optimal solution to a problem quickly.

Meanwhile, with its powerful image processing capabilities, CNN can accurately capture and identify complex financial data patterns. Combining the two enables this optimization method to provide enterprises with accurate financial accounting and forecasting based on a deep understanding of the complex business environment. By harnessing the capabilities of artificial intelligence and advanced algorithms, our intelligent accounting method aims to deliver more precise and reliable financial information. This will empower management to make informed decisions, enhancing enterprise competitiveness in today’s competitive business landscape.

Related Works

The intelligent accounting optimization method, integrating metaheuristic algorithms and CNN, combines the search and parameter optimization capabilities of metaheuristic algorithms with the feature extraction prowess of CNN to enhance intelligent accounting processes. This approach enables identifying optimal or near-optimal solutions within extensive search spaces, enhancing the system’s ability to handle complex accounting problems autonomously. Automating decision-making processes reduces intervention requirements and increases accounting accuracy. This method aims to imbue accounting systems with greater intelligence, adaptability, and efficiency, providing enterprises with more precise and effective financial decision-making capabilities.

In recent years, various optimization methods in intelligent accounting have significantly enhanced the efficiency of enterprise financial management. Kim et al. (2023) combined cluster analysis with Data Envelopment Analysis (DEA) to analyze core competitiveness positioning and optimise accounting practices. Su & Sun (2023) evaluated enterprise performance based on accounting data, offering critical insights for investors, senior managers, and government decision-makers. Bose, Dey & Bhattacharjee (2023) addressed profit estimation inaccuracies by designing performance evaluation indicators linked to accounting data, reducing operational profit errors. Li, Wang & Yang (2023) focused on improving enterprise management and performance systems by analyzing traditional accounting method issues and proposing structural enhancements using BP neural networks. Brusov & Filatova (2023) applied linear analysis methods to explore accounting system impacts, while Chang, Lee & Lee (2023) conducted correlation analyses using profitability and relative value indicators alongside corresponding data. Oyedele et al. (2023) utilized financial indicators across profitability, operational capacity, cash flow, debt-paying ability, and growth potential dimensions to construct a performance evaluation model, employing DEA for comprehensive ranking calculations based on listed company data.

Metaheuristics are algorithms that emulate the optimization processes observed in natural or social systems. These algorithms generally operate independent of specific problem structures, relying instead on heuristic rules inspired by natural selection, physics, or analogous phenomena. Their primary objective is to discover optimal or near-optimal solutions within a problem’s solution space, iteratively refining candidate solutions. Fontes, Homayouni & Goncalves (2023) utilized the simulated annealing algorithm as a local search method and employed particle swarm optimization to tackle integrated production and transportation job scheduling under vehicle constraints. Defersha, Obimuyiwa & Yimer (2022) formulated a mathematical framework for flexible job resource scheduling, considering order-dependent constraints and integrating the simulated annealing algorithm. Wang et al. (2022b) addressed the scheduling of distributed pipelines with sequence-dependent constraints, proposing a two-stage iterative greedy approach. Sun et al. (2023) combined genetic algorithms and variable neighbourhood search to optimize scheduling, emphasizing load balancing through customized crossover and mutation operators. Xu & Yu (2022) introduced a dynamic shrinkage method using adaptive trade-off and evolution strategies to enhance efficiency and meet load requirements. Abido & Elazouni (2021) developed a multi-objective evolutionary approach for project scheduling involving multi-mode activities to minimize time and capital costs. Pan et al. (2021) presented a deep reinforcement learning-based optimization method for scheduling problems, employing convolutional neural networks to generate end-to-end feasible solutions trained with actor-critic methods and refined via metaheuristics. Song et al. (2022) utilized graph neural networks to learn priority scheduling rules, leveraging a heterogeneous architecture for offline training and rule relationship capture. Tian, Liu & Lv (2024) proposed a new nature-inspired meta-heuristic algorithm, the Snow Goose Algorithm, which mimics the flight patterns observed during their migratory process. Jia, Lu & Xing (2024) introduced an evolutionary updating mechanism called the Memory Backtracking Strategy, which comprises the thinking, recall, and memory phases. Liang et al. (2024) made several adjustments to the original Ant Colony Optimization algorithm to make it effective for the problem of maximizing influence.

Intelligent Accounting Optimization Method Based on Meta-Heuristic and CNN

To enhance corporate accounting operations and provide enterprises with high-quality financial data and decision-making advice, we propose an advanced method using meta-heuristics and convolutional neural networks (CNNs). This approach aims to analyze enterprise accounting intelligently, encompassing multi-modal accounting feature extraction (MAFE) using CNN, accounting quality assessment (AQAA), and meta-heuristics-based accounting optimization (MAO).

Figure 1 illustrates our technology roadmap. First, we introduce a CNN-based MAFE method to address accounting data’s complex, multi-modal nature. Traditional methods often struggle to extract features from diverse financial and non-financial information comprehensively. CNN’s robust feature learning capability enables the extraction of key features from numerical data and unstructured sources like text and images, providing critical support for subsequent AQAA and method optimization.

10.7717/peerjcs.2281/fig-1 Figure 1 The technology roadmap of our method.

Next, we propose an AQAA method to evaluate the quality of extracted accounting features. This involves constructing a machine learning-based evaluation model to assess feature accuracy and reliability objectively. By identifying accounting issues promptly, this method guides the optimization of accounting practices.

Finally, we present a meta-heuristics-based MAO method for optimizing accounting methodologies. Meta-heuristics, known for their global optimization prowess, navigate complex search spaces to find optimal solutions. Integrating these algorithms with accounting methodologies allows for intelligent, self-adaptive parameter and strategy optimization during the accounting process. This adaptive approach enhances efficiency and enables automatic adjustment of accounting methods to meet varying enterprise requirements.

In summary, the intelligent accounting optimization method based on meta-heuristics and CNN proposed in this article realizes the intelligence and efficiency of accounting through three steps: multi-modal accounting feature extraction, accounting quality assessment and accounting method optimization. This method not only improves the accuracy of accounting but also provides high-quality financial data and decision-making suggestions for enterprises, which can help enterprises better respond to market changes and challenges.

CNN-based multi-modal accounting feature extraction

We propose a CNN-based multi-modal accounting feature extraction method to deepen our understanding of accounting information and enhance method accuracy. Accounting data, comprising raw numbers, letters, and special symbols, is structured as “documents, vouchers, accounts, statements” during accounting events. These records capture objective observations and essential facts, utilizing distinct symbols to differentiate and identify various types of information. These symbols encompass quantitative numerical data and qualitative non-numerical data, forming the foundation of accounting information. Therefore, this study focuses on extracting multi-modal features from accounting information, encompassing commonly used documents, vouchers, and statements in accounting practices, as illustrated in Fig. 2.

10.7717/peerjcs.2281/fig-2 Figure 2 Multi-modal accounting information.

In the accounting data processing workflow, accounting data primarily refers to the original data reflecting the increase and decrease in funds during the execution of production and business activities or budget plans. It also covers a series of objective facts that do not directly involve changes in funds but must be recorded and reflected in accounting. These data originate from diverse sources and exhibit various types, characterized by their systematic, periodic, and continuous nature and their high reuse value. Given these unique properties, this article employs an optimized CNN technique to extract features from accounting data. Using CNN, valuable feature information is extracted from vast amounts of accounting data. These feature representations can reflect the complex structure and inherent relationships within accounting data, providing robust support for subsequent accounting analysis and forecasting. This article integrates a location enhancement module to explicitly model each textual element to enhance the even treatment of data sections and contextual backgrounds in current methods.The sequence of attention maps generated by the text detection recognizer is denoted by hattn ∈ ℝT×H×W, where T is the maximum sequence length. Since the text length of different accounting images differs, the original attention map may be misleading due to attention drift. Therefore, this article adopts compression and expansion strategies to solve these problems. Denote the length of a character sequence as L, which hattnj represents the attention map corresponding to the j-th text character. Since the text length varies from image to image, hattnjj=1L can be obtained by selecting the previous L effective attention map. Then, concatenate them, using a maximum operator to reduce the channel dimension to 1. The result is denoted by hscore: (1) hscore=MaxConcathattn1,…,hattnL.

Use the C convolution kernel to upsample hscore, and use the Softmax function to obtain the result hpos: (2) hpos=SoftmaxConvhscore.

The mean and variance contain the information of the image. To facilitate the subsequent registration between the image and the text, we perform instance normalization for XI, to make the focus of the image more prominent. The normalized features are multiplied with hpos to obtain position-enhanced features Xpos: (3) Xpos=INXI⨂hpos

The key to the fusion of visual and semantic features is the registration of character regions rather than the attention to background, so feature selection technology is crucial. Instead of manual labelling and individual scoring methods, we can consider using easy-to-obtain attention maps for feature selection. Since pixel-wise character confidence is included in hscore, the one with the highest score in hscore is selected, resulting in the foreground coordinate set F: (4) F:=m,n:hscorem,n∈TopKhscore.

Next, we use the coordinate set F as an index to collect the foreground character features in Xpos. To avoid loss of neighborhood information, a pixel-to-region strategy is used before collection, where each index pixel is representative of its neighbourhood by weighted summation of local regions: (5) Xpos′= ∑△m,△n∈Nω△m,△nXposm+△m,n+△n.

where N represents the neighborhood displacement and ω(⋅, ⋅) represents the weight of each displacement. Finally, the coordinate pair in F can index Xpos′ to obtain the selected feature XS=Xpos′i,ji,j∈F.

When dealing with documents and vouchers in accounting information, we implement multimodal information extraction by embedding keyword fuzzy matching and the Levenshtein distance algorithm. After extracting information from different modalities, such as text, tables, and images, the system needs to integrate these pieces of information. During the integration process, it is crucial to ensure the accuracy and completeness of the information while also considering the correlations and logical relationships between different modalities. We further embed sequential reconstruction methods for information extraction for complex or specialised accounting documents and vouchers.

Accounting quality assessment methods for accounting

The accounting quality evaluation method for accounting can further evaluate the accounting quality of enterprises by using the obtained features XS, and provide standards for improving the accounting ability of enterprises.

To solve the problems of local minimization and over-fitting of multi-modal features XS. We use the Adaboost ensemble learning model to optimize the evaluation results for the accounting objectives of enterprise accounting. The model performs the regression task by weighted combination of multiple weak learners. Firstly, a series of individual learners are generated, and then a suitable combination strategy is used to integrate them into a stronger learner, as shown in Fig. 3.

10.7717/peerjcs.2281/fig-3 Figure 3 Adaboost ensemble model based on CNN.

We employ a sequential structure to link multiple homogeneous weak classifiers, enhancing classifier stability and accuracy. The final learning outcome hinges on effectively combining the classification results from these weak classifiers. Central to our approach is the method for generating diverse weak classifiers. We achieve this by iteratively adjusting the sample weights and increasing focus on previously misclassified samples during subsequent training iterations. After completing all iterations, weights are assigned to each weak classifier based on its classification error rate, ensuring those with lower errors receive greater weights. Ultimately, these weighted weak classifiers are aggregated to form a robust learner with improved predictive accuracy.

If the data quality feature training set sample is X, and its corresponding quality class is w. The weight distribution for high-quality data is initialized, assigning equal weights wi to each training sample. Perform m iterations, train each weak classifier in turn, select the weights of the weak classifier decision tree Bm(x) to train the quality feature data, and calculate the error rate em of the weak classifier at each iteration: (6) em=∑N=1Qwim∣ci≠Bxi∑N=1Qwim.

Calculate the weights αm for the weak classifier Bm(x): (7) αm=12ln1−emem.

Iterate until m is equal to the set number of iterations M, and the final strong classifier G(x) is obtained by weighting the sum of each weak classifier: (8) Gx= sin∑m=1Mα mBmx.

Adaboost can not only effectively deal with the local minimization problem caused by multi-modal features but also alleviate the over-fitting phenomenon to a certain extent, to provide more accurate and stable evaluation results for the accounting goal of enterprise accounting XS.

Meta-heuristics-based accounting optimization method

Accounting optimization through meta-heuristics aims to enhance enterprise efficiency in accounting processes. These algorithms integrate meta-heuristic principles and optimization techniques to tackle the diverse and complex challenges inherent in enterprise accounting. This study optimises accounting methodologies by leveraging multi-modal features and quality assessment outcomes.

The approach employs meta-heuristic algorithms such as Genetic Algorithm (GA) and Simulated Annealing (SA) to solve accounting models and determine optimal enterprise operations. GA, renowned for addressing intricate optimization problems, operates with three key functions—selection, mutation, and crossover—each utilizing parameters to refine solutions. Such constraint is solved by maximizing the accounting quality assessment value, namely: (9) maxE= ∑i=1nc×x j+xi2+B2n×DL

(10) ∑i=1mX i=L

where L refers to the cost of the whole accounting process, each xi denotes the GA variable and xj represents the SA variable.

Mutation function is an important part of GA, and it includes Gaussian variograms, uniform variograms, and adaptive variograms. We chose to use a Gaussian variator with a shrinkage scale factor to successfully tune the minimum energy. When the shrinkage parameter is 1, the variance linearly decreases to 0 over the generations. This design makes the mutation process more flexible and controllable, which is helpful in boosting the performance of GA.

The SA model is locally based on the solution of a constrained optimization model. This constrained optimization model has the following steps:

(1) The corresponding objective gj+1 is randomly generated by providing the starting temperature and selecting the variable xj. The temperature then decreases as the number of cycles increases.

(2) Each time through the loop, a novel objective gj+1 is created and compared with the current objective.

(3) If gj+1<gj, replace g j with gj+1 and enter the next cycle until the temperature value is really reduced.

The key of the SA model is to escape from the local optimal solution, which is determined by the Metropolis criterion. At high temperatures, the probability of which is wrong solution is high. As the temperature decreases, the probability of accepting the suboptimal solution is gradually reduced so the algorithm tends to search near the optimal solution.

The accounting optimization method based on Yuanqifa considers the diversity and complexity of enterprise accounting problems more comprehensively and is committed to providing more effective accounting solutions. This method can be popularized and applied in different industries and enterprise types to promote the improvement of enterprise accounting efficiency.

Experiment and Analysis

Dataset and implement details

We tested intelligent accounting using a method combining CNN and meta-heuristics optimization, utilizing the Synthetic Financial Dataset available at Zenodo (DOI: 10.5281/zenodo.7543591).

Our training setup utilized hardware comprising an i5-14400F processor and 2 Nvidia RTX 4060Ti GPUs. Throughout the training phase, TensorFlow was our deep learning framework, configured with specific settings detailed in Table 1.

10.7717/peerjcs.2281/table-1 Table 1 Implementation parameters.

Parameters	Value	
Initial learning rate	9 × 10−4	
Epoch	600	
Batch-size	60	
Decay	0.92	
Gradient descent method	Adam	

During model training, we incorporated a weight decay term with a value of 0.00009. These choices in test environments and configurations were made to ensure our method demonstrates superior efficiency and accuracy in addressing intelligent accounting challenges.

In this article, Accuracy (Acc) and Mean Average Precision (mAP) are employed as the criteria of intelligent accounting optimization method based on meta-heuristics and CNN, and the calculation formula is as follows: (11) Acc=YprYgt

(12) mAP=∑APin

where pr and gt refer to the predicted and true values, n denotes the entire classes, and APi is the average precision for the i th class. AP is the area under the precision and recall curves and reflects the model’s performance in each class. Therefore, the mAP means the average of the AP over all classes and is used to measure the average performance of the model over all classes. In addition, when calculating AP, we first need to get the precision and recall values.

Ablation experiments

Table 2 presents detailed ablation experiments on intelligent accounting optimization methods using meta-heuristic algorithms and CNNs based on the Synthetic Financial Dataset. We systematically evaluate the impact of three modules: MAFE, AQAA, and MAO, both individually and in various combinations, on model performance.

10.7717/peerjcs.2281/table-2 Table 2 Ablation experiments.

MAFE	AQAA	MAO	mAP@0.5	mAP@0.75	mAP@0.95	mAP@0.5:0.95	
			0.755	0.732	0.701	0.733	
O			0.779	0.752	0.739	0.753	
	O		0.766	0.745	0.731	0.749	
		O	0.786	0.754	0.748	0.767	
O	O		0.799	0.779	0.768	0.785	
O		O	0.812	0.801	0.783	0.803	
	O	O	0.824	0.809	0.798	0.815	
O	O	O	0.852	0.834	0.812	0.838	

Firstly, individually introducing MAFE enhances mAP@0.5:0.95 to 0.739, AQAA contributes a score of 0.731 for mAP@0.5:0.95, and MAO achieves 0.748 for mAP@0.5:0.95. These results highlight the positive influence of each module in optimizing accounting processes.

Next, through detailed analysis in Figs. 4 and 5, we observe that combining these modules consistently improves model performance. Specific combinations demonstrate significant achievements, with some configurations reaching as high as 0.798 on mAP@0.95, underscoring the synergy between modules.

10.7717/peerjcs.2281/fig-4 Figure 4 Ablation experiments results in term of mAP with single module.

10.7717/peerjcs.2281/fig-5 Figure 5 Ablation experiments result in terms of mAP with different modules.

Finally, integrating all three modules—MAFE, AQAA, and MAO—yields compelling results: mAP@0.5 scores 0.852, mAP@0.75 scores 0.834, and mAP@0.95 scores 0.812. The overall mAP@0.5:0.95 reaches 0.838, affirming the effectiveness of our method and demonstrating the potential of multi-module combinations to enhance model performance in intelligent accounting optimization.

Our ablation experiments comprehensively explore the roles of MAFE, AQAA, and MAO modules in enhancing intelligent accounting optimization. These findings validate the efficacy of our approach and offer promising avenues for future research in this field.

Compare our method and other methods

In our experiment on the Synthetic Financial Dataset, we evaluated the performance of the MAFE method by comparing it with DAMUN (Feng et al., 2023), FDGNet (Zhang et al., 2023), AdaMoW (Zhang, Wu & Huang, 2023), and Megcf (Liu et al., 2023a). The results, detailed in Fig. 6 and Table 3, demonstrate that MAFE outperforms all other methods across all evaluation metrics.

10.7717/peerjcs.2281/fig-6 Figure 6 Comparison of MAFE with other extraction methods.

Specifically, MAFE achieved remarkable results across various accuracy metrics, with scores of 0.887 for Acc@1, 0.902 for Acc@5, 0.936 for Acc@10, and 0.966 for Acc@20. Compared to DAMUN, our method significantly improved 3.2% in Acc@1. Furthermore, MAFE surpassed FDGNet and AdaMoW in all indicators, outperforming them by 2.3% in Acc@5 and 1.4% in Acc@10, respectively. Lastly, MAFE achieved a 3.0% lead over Megcf in Acc@20. These comparisons underscore MAFE’s superior performance in intelligent accounting. MAFE excels in accuracy, ranking first in Acc@1, and demonstrates significant advancements in broader recommendation accuracy and scope. MAFE’s superiority equips enterprises with more reliable and efficient intelligent solutions to address increasingly complex financial data and accounting challenges.

In our detailed analysis of AQAA’s performance on the Synthetic Financial Dataset, we compared it against well-known methods CRFIQA (Boutros et al., 2023), PCQA (Liu et al., 2023b), and FAQA (Idland et al., 2023) to thoroughly assess its strengths and capabilities. As shown in Fig. 7 and Table 4, the results highlight AQAA’s significant advantages across several key evaluation metrics.

Compared to CRFIQA, AQAA excels in overall performance and leads by a significant 2.8% in the Acc@5 metric, exemplifying its prowess in precisely categorizing questions. Furthermore, when benchmarked against PCQA and FAQA, AQAA maintains its competitive edge. Specifically, AQAA outperforms PCQA with a remarkable 6.3% improvement in mAP@0.75, demonstrating its robustness in accurately identifying problematic areas. Additionally, AQAA surpasses FAQA with a 1.8% enhancement in Acc@5, further reinforcing its effectiveness in enhancing the accuracy of question identification. These findings collectively validate AQAA’s prowess in improving the precision and effectiveness of question identification within the context of intelligent accounting optimization.

Through our detailed examination of MAO’s performance on the Synthetic Financial Dataset, we have successfully enhanced AQAA’s capabilities through parameter optimization. Comparing MAO with established benchmarks CRFIQA, PCQA, and FAQA, we observe that MAO achieves superior performance in Acc@1 and mAP@0.95 values, reaching 0.943 and 0.812, respectively. These detailed results are illustrated in Fig. 8.

10.7717/peerjcs.2281/table-3 Table 3 Comparison of MAFE with other methods.

Methods	Acc@1	Acc@5	Acc@10	Acc@20	
DAMUN	0.855	0.887	0.904	0.922	
FDGNet	0.846	0.879	0.892	0.914	
AdaMoW	0.857	0.888	0.897	0.929	
Megcf	0.851	0.891	0.911	0.936	
Ours	0.887	0.902	0.936	0.966	

10.7717/peerjcs.2281/fig-7 Figure 7 The results of our AQAA.

10.7717/peerjcs.2281/table-4 Table 4 Comparison of AQAA with other methods.

Methods	Acc@1	Acc@5	mAP@0.75	mAP@0.5:0.95	
CRFIQA	0.897	0.916	0.736	0.757	
PCQA	0.874	0.896	0.716	0.741	
FAQA	0.904	0.926	0.731	0.749	
Ours	0.921	0.944	0.779	0.785	

10.7717/peerjcs.2281/fig-8 Figure 8 Comparison of MAO with other methods.

MAO’s optimization approach stands out among other advanced methods, exhibiting remarkable performance in accounting optimization. These superior metrics offer enterprises more efficient and dependable solutions for refining their accounting processes. The significant improvements in Acc@1 and mAP@0.95 values demonstrate MAO’s robust adaptability in handling intricate financial data and accounting challenges. As a result, enterprises can confidently rely on MAO to provide flexible and effective accounting solutions, which are crucial for navigating dynamic economic environments and regulatory landscapes. MAO’s method not only achieves substantial enhancements over existing benchmarks but also showcases its versatility in diverse domains and complex scenarios. This comprehensive performance validates MAO’s excellence in intelligent accounting, offering enterprises a reliable basis for decision-making and investors credible financial information.

Finally, to simulate the performance of our method in real-world scenarios, as shown in Fig. 9, we restructured the data to form tourism sample sets with varying numbers of real-world samples, including 20, 50, 100, 200, and 500 samples. By comparing the running time under different sample sets, we better understood our method’s performance with different data scales. As shown in Fig. 7, we have recorded the computation time of MAFE, AQAA, and MAO under different sample sizes. Comparative analysis shows that regardless of the actual scenario, the system’s computation time remains below 210 ms. This result indicates that our model can handle concurrent requests that may arise in real-world scenarios, thus providing users with stable and efficient services.

10.7717/peerjcs.2281/fig-9 Figure 9 System response efficiency test.

Discussion

The experiments conducted on CNN-based multi-modal accounting feature extraction, accounting quality assessment, and meta-heuristics-based accounting optimization have yielded satisfactory results, validating the effectiveness of intelligent accounting optimization using meta-heuristics and CNN. This approach demonstrates exceptional performance across various environmental and data challenges, offering a comprehensive and practical solution for accounting tasks.

The CNN-based multi-modal accounting feature extraction technique successfully extracts crucial information from complex accounting data. By integrating data from diverse modalities, our method comprehensively reflects enterprise financial statuses and business outcomes, providing a robust foundation for subsequent accounting quality assessment and optimization. Secondly, our adopted method for accounting quality assessment incorporates deep data mining and machine learning algorithms, enhancing the objectivity and accuracy of evaluation results beyond traditional financial indicators. This capability aids enterprises in the timely identification of risks and issues, facilitating informed decision-making. Lastly, the meta-heuristics-based accounting optimization method optimizes accounting processes by intelligently adjusting model parameters and structures. This not only boosts accounting accuracy and efficiency but also reduces manual intervention, thereby easing the workload on financial personnel.

In summary, our intelligent accounting optimization method leveraging meta-heuristics and CNN proves to be an efficient, reliable solution with broad application potential. We aim to explore additional optimization strategies and technological advancements to enhance accounting accuracy further and support enterprise development.

Conclusion

We introduce an innovative intelligent accounting optimization method combining metaheuristic algorithms and CNN to enhance enterprise accounting capabilities. Initially, we employ a CNN-based multi-modal accounting feature extraction approach to derive multi-modal representations from accounting data, encompassing financial documents, vouchers, and related information. Building upon this foundation, we develop an accounting quality evaluation method and establish a comprehensive accounting evaluation system. Leveraging these multi-modal features and the evaluation system, we integrate a meta-heuristic-based accounting optimization method to enhance accounting processes intelligently. Experimental results demonstrate significant achievements, including an accuracy of 0.943 and an mAP score of 0.812. These outcomes validate our method as an effective technical solution for elevating accounting quality. In the future, we will continue to deepen and expand this intelligent accounting optimization approach. First, we plan to optimize the CNN model further to extract more refined and comprehensive accounting data features. By introducing more advanced network structures and algorithms, we can capture more subtle financial changes, thereby enhancing the accuracy and efficiency of accounting analysis. Moreover, we will explore how to integrate meta-heuristic algorithms with CNN better to achieve more intelligent and efficient accounting optimization.

Supplemental Information

10.7717/peerj-cs.2281/supp-1 Supplemental Information 1 Code

Additional Information and Declarations

Competing Interests

Author Contributions

Data Availability

The author declares there are no competing interests.

Yanrui Dong conceived and designed the experiments, performed the experiments, analyzed the data, performed the computation work, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.

The following information was supplied regarding data availability:

The code is available in the Supplementary File.

The dataset is available at Zenodo: Mavsar, M. (2022). Video-Trajectory Robot Dataset [Data set]. Zenodo. https://doi.org/10.5281/zenodo.6337847.
==== Refs
References

Abido & Elazouni (2021) Abido MA Elazouni A 2021 Modified multi-objective evolutionary programming algorithm for solving project scheduling problems Expert Systems with Applications 183 115338 10.1016/j.eswa.2021.115338
Alazzazbi, Mustafa & Karage (2023) Alazzabi WYE Mustafa H Karage AI Risk management, top management support, internal audit activities and fraud mitigation Journal of Financial Crime 2023 30 2 569 582
Berdiyeva, Islam & Saeedi (2021) Berdiyeva O Islam MU Saeedi M 2021 Artificial intelligence in accounting and finance: meta-analysis International Business Review 3 1 56 79
Bose, Dey & Bhattacharjee (2023) Bose S Dey SK Bhattacharjee S 2023 Big data, data analytics and artificial intelligence in accounting: an overview Handbook of big data research methods
Boutros et al. (2023) Boutros F Fang M Klemt M Fu B Damer N 2023 CR-FIQA: face image quality assessment by learning sample relative classifiability Proceedings of the IEEE/CVF conference on computer vision and pattern recognition Piscataway IEEE 5836 5845
Brusov & Filatova (2023) Brusov P Filatova T 2023 Capital structure theory: past, present, future Mathematics 11 3 616 10.3390/math11030616
Chang, Lee & Lee (2023) Chang A Lee TS Lee HM 2023 Applying sustainable development goals in financial forecasting using machine learning techniques Corporate social responsibility and environmental management
Chen & Chao (2022) Chen C Chao Y 2022 Intelligent statistical method of accounting information teaching data based on SVM International conference on e-learning, e-education, and online training Springer Nature Switzerland Cham 589 600
Chen & Metawa (2020) Chen X Metawa N 2020 Enterprise financial management information system based on cloud computing in big data environment Journal of Intelligent & Fuzzy Systems 39 4 5223 5232 10.3233/JIFS-189007
Cui & Verma (2022) Cui P Verma V 2022 An intelligent accounting system based on data mining algorithm International conference on cognitive based information processing and applications Springer Nature Singapore Singapore 249 258
Defersha, Obimuyiwa & Yimer (2022) Defersha FM Obimuyiwa D Yimer AD 2022 Mathematical model and simulated annealing algorithm for setup operator constrained flexible job shop scheduling problem Computers & Industrial Engineering 171 108487 10.1016/j.cie.2022.108487
Desyatnyuk, Muravskyi & Shevchuk (2021) Desyatnyuk O Muravskyi V Shevchuk O 2021 Accounting automation in agroindustrial enterprises using drones (UAVs) 2021 11th international conference on advanced computer information technologies (ACIT) Piscataway IEEE 337 341
Ding (2022) Ding M 2022 Construction of enterprise financial information intelligent processing innovation model based on internet of things technology Computational Intelligence and Neuroscience
Feng et al. (2023) Feng X Weng Y Li W Chen P Zheng H 2023 DAMUN: a domain adaptive human activity recognition network based on multimodal feature fusion IEEE Sensors Journal 23 18 22019 22030 10.1109/JSEN.2023.3300357
Fontes, Homayouni & Goncalves (2023) Fontes DBMM Homayouni SM Goncalves JF 2023 A hybrid particle swarm optimization and simulated annealing algorithm for the job shop scheduling problem with transport resources European Journal of Operational Research 306 3 1140 1157 10.1016/j.ejor.2022.09.006
Guo (2021) Guo Y 2021 CNS: interactive intelligent analysis of financial management software based on apriori data mining algorithm International Journal of Cooperative Information Systems 30 01n04 2150008 10.1142/S0218843021500088
Hernes & Sobieska-Karpińska (2016) Hernes M Sobieska-Karpińska J 2016 Application of the consensus method ina multiagent financial decision support system Information Systems and e-Business Management 14 167 185 10.1007/s10257-015-0280-9
Hu (2022) Hu Y 2022 Enterprise intelligent accounting system structure and intelligent accounting algorithm 2021 international conference on big data analytics for cyber-physical system in Smart City: volume 1 Springer Singapore 1293 1298
Idland et al. (2023) Idland S Kramer-Johansen J Bakke HK Hjortdahl M 2023 Assessing bystander first aid: development and validation of a First Aid Quality Assessment (FAQA) tool BMC Emergency Medicine 23 39 10.1186/s12873-023-00811-z 37013526
Jia, Lu & Xing (2024) Jia H Lu C Xing Z 2024 Memory backtracking strategy: an evolutionary updating mechanism for meta-heuristic algorithms Swarm and Evolutionary Computation 84 101456 10.1016/j.swevo.2023.101456
Kim et al. (2023) Kim NH He F Zhang H Hong KR Ri KC 2023 A data envelopment analysis-based clustering approach under dynamic situations European Journal of Operational Research 311 1 251 262 10.1016/j.ejor.2023.04.032
Lai (2022) Lai M 2022 Smart financial management system based on data ming and man-machine management Wireless Communications and Mobile Computing 2022 2717982
Li Cui (2022) Li Cui Y 2022 Research on the construction of intelligent accounting personnel training system under digital economy 2022 5th international conference on humanities education and social sciences (ICHESS 2022) Dordrecht Atlantis Press 2973 2977
Li, Wang & Yang (2023) Li X Wang J Yang C 2023 Risk prediction in financial management of listed companies based on optimized BP neural network under digital economy Neural Computing and Applications 35 3 2045 2058 10.1007/s00521-022-07377-0
Liang et al. (2024) Liang JC Gong YJ Wu XK Li Y 2024 Customized influence maximization in attributed social networks: heuristic and meta-heuristic algorithms Complex & Intelligent Systems 10 1 1409 1424 10.1007/s40747-023-01220-2
Liu et al. (2023a) Liu K Xue F Guo D Wu L Li S Hong R 2023a Megcf: multimodal entity graph collaborative filtering for personalized recommendation ACM Transactions on Information Systems 41 2 1 27 10.1145/3544106
Liu et al. (2023b) Liu Y Yang Q Xu Y Yang L 2023b Point cloud quality assessment: dataset construction and learning-based no-reference metric ACM Transactions on Multimedia Computing, Communications and Applications 19 2S 1 26 10.1145/3550274
Mancini, Lombardi & Tavana (2021) Mancini D Lombardi R Tavana M 2021 Four research pathways for understanding the role of smart technologies in accounting Meditari Accountancy Research 29 5 1041 1062 10.1108/MEDAR-03-2021-1258
Oyedele et al. (2023) Oyedele AA Ajayi AO Oyedele LO Bello SA Jimoh KO 2023 Performance evaluation of deep learning and boosted trees for cryptocurrency closing price prediction Expert Systems with Applications 213 119233 10.1016/j.eswa.2022.119233
Pan et al. (2021) Pan Z Wang L Wang J Lu J 2021 Deep reinforcement learning based optimization algorithm for permutation flow-shop scheduling IEEE Transactions on Emerging Topics in Computational Intelligence 7 4 983 994 10.1109/TETCI.2021.3098354
Qu (2023) Qu H 2023 Design of enterprise intelligent accounting system structure and intelligent accounting algorithm International conference on mathematics, modeling, and computer science (MMCS2022), vol. 12625 Bellingham SPIE 352 357
Song et al. (2022) Song W Chen X Li Q Cao Z 2022 Flexible job-shop scheduling via graph neural network and deep reinforcement learning IEEE Transactions on Industrial Informatics 19 2 1600 1610 10.1109/TII.2022.3189725
Su & Sun (2023) Su J Sun Y 2023 An improved TOPSIS model based on cumulative prospect theory: application to ESG performance evaluation of state-owned mining enterprises Sustainability 15 13 10046 10.3390/su151310046
Sun et al. (2023) Sun K Zheng D Song H Cheng Z Lang X Yuan W Wang J 2023 Hybrid genetic algorithm with variable neighborhood search for flexible job shop scheduling problem in a machining system Expert Systems with Applications 215 119359 10.1016/j.eswa.2022.119359
Tian, Liu & Lv (2024) Tian AQ Liu FF Lv HX 2024 Snow Geese Algorithm: a novel migration-inspired meta-heuristic algorithm for constrained engineering optimization problems Applied Mathematical Modelling 126 327 347 10.1016/j.apm.2023.10.045
Wang et al. (2022a) Wang Z He Y Jiang H Yu C 2022a Enterprise intelligent accounting system structure and intelligent accounting algorithm International conference on frontier computing Springer Nature Singapore Singapore 1085 1092
Wang et al. (2022b) Wang ZY Pan QK Gao L Wang YL 2022b An effective two-stage iterated greedy algorithm to minimize total tardiness for the distributed flowshop group scheduling problem Swarm and Evolutionary Computation 74 101143 10.1016/j.swevo.2022.101143
Wang, Zhang & Tao (2021) Wang X Zhang Y Tao Y 2021 Research on system construction and application of enterprise intelligent finance from the perspective of artificial intelligence 2021 international conference on computer information science and artificial intelligence (CISAI) Piscataway IEEE 597 602
Wu (2021) Wu Y 2021 Intelligent learning of digital finance and accounting based on neural network technology 2021 International conference on aviation safety and information technology 44 48
Xu & Yu (2022) Xu WY Yu X 2022 A novel space contraction based on evolutionary strategy for economic dispatch Knowledge-Based Systems 237 107743 10.1016/j.knosys.2021.107743
Yuan (2023) Yuan H 2023 Research on the application of computer intelligent technology in cost accounting and financial management International conference on frontier computing Springer Nature Singapore Singapore 217 222
Zhang et al. (2023) Zhang G Nie R Cao J Chen L Zhu Y 2023 FDGNet: a pair feature difference guided network for multimodal medical image fusion Biomedical Signal Processing and Control 81 104545 10.1016/j.bspc.2022.104545
Zhang, Wu & Huang (2023) Zhang J Wu X Huang C 2023 AdaMoW: multimodal sentiment analysis based on adaptive modality-specific weight fusion network IEEE Access 11 48410 48420 10.1109/ACCESS.2023.3276932
Zhou (2022) Zhou Y 2022 Intelligent accounting system based on data mining algorithm 2022 fourth international conference on emerging research in electronics, computer science and technology (ICERECT) Piscataway IEEE 1 4
Zhuo (2022) Zhuo J 2022 Analysis of enterprise financial management under the trend of intelligent accounting Learning & Education 10 9 41 43
