optimization algorithms

Sometimes companies file fraudulent financial statements for tax fraud. The purpose of this research is to combine data mining tools and artificial intelligence with meta-heuristic algorithms to explain and optimize a model for detecting fraud and tax evasion by using the financial reporting capacity of companies. In order to collect research data, the financial reports of companies listed on the Tehran Stock Exchange, including 1056 years-company in the years 2007 to 2020 were examined. First, the effect of qualitative and quantitative indicators derived from financial reports was investigated, refined and then used to form the explaining model in the Adaptive Neural-Fuzzy Inference System (ANFIS). The model was combined and developed in the optimization stage with the aim of maximizing detection power, by using meta-heuristic optimization algorithms including genetic algorithm, particle swarm optimization algorithm and differential evolution algorithm. The research findings show that in terms of predictive power among the optimization algorithms used in the research, the particle swarm algorithm with the highest percentage of correct prediction, produces the most optimal model and in examination by test and training data is the most efficient algorithm. The results of the research indicate that the use of different optimization algorithms in the data mining approach increases the predictive power of the fraudulent fiscal reporting model.

Transportation system brings the problem of vehicle routing that seeks optimal allocation of customers to vehicles in distribution system. There are various types of vehicle routing problems that try to provide models that are closer to the real world constraints by considering limitations such as vehicle capacity, number of demands, distance and so forth. The transportation network of Digikala Company inside and between cities is complicated based on the number of available routes, variety of means of transport and multiple distribution centers in the provincial areas. Digikala's distribution network is not only a geometric representation of the transport routes, but also represents the transport system. In this research, studying variables of shipping these goods has been considered, firstly, to identify a mathematical model for the distribution of these goods and then to examine how to solve this mathematical model, which is estimative and accurate. We solved this problem after examining and proving the model solution through approximate ways using meta-heuristic method which gives us the closest answers in NP-hard problems. Finally, while coding in small-scale Matlab software, we study the distribution model of the company and obtain the required output by proving the downward trend of the final function. The study confirms that online store companies can design, optimize, and manage complex supply chain operations at different levels with the help of Genetic Algorithm