Intelligent optimization and control system for accounting cost calculation based on reinforcement learning

Engineering notes

The proposed system outperforms other approaches, achieving lower MAPE (3.600%) and training time (6.25 s).

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Modern cost accounting needs to be more adaptive and intelligent with the increasing speed and volume of financial operations and digital transactions. Conventional approaches are based on fixed rules and manual tweaks, which cannot accommodate dynamic business needs. The cost accounting and budget overrun data contain 2643 samples. Preprocessing steps include min–max normalization of heterogeneous financial indicators and missing value imputation to handle incomplete or noisy records. A principal component analysis (PCA)-based feature extraction module reduces dimensionality and identifies critical cost-driving factors, enhancing learning efficiency. This research proposes an intelligent optimization and control system for accounting cost management using dynamic honey badger-twin delayed deep deterministic policy gradient (DHB-twin delayed DDPG) using Python, which ensures stable and efficient learning in complex financial environments. Optimize cost allocation and prediction while dynamically adapting to changing accounting scenarios, marking a novel integration of DHB-inspired exploration with twin delayed DDPG for financial decision-making. By modeling cost management as a sequential decision-making problem, the DHB-twin delayed DDPG agent learns optimal actions such as budget adjustment and resource allocation through interaction with the dynamic accounting environment. The proposed system outperforms other approaches, achieving lower MAPE (3.600%) and training time (6.25 s). The research demonstrates that combining advanced models enables a dynamic, intelligent, and efficient cost accounting system, providing a significant step toward fully autonomous financial decision support.

Links and sources

Comments