1. 基于改进的五层模糊神经网络 的农业科技成果估值研究.
- Author
-
曹艳, 刘强, 胡亮, 胡旭, and 刘远利
- Subjects
- *
FUZZY neural networks , *AGRICULTURAL technology , *MACHINE learning , *ACADEMIC achievement - Abstract
[Objective] To solve the problem of difficult valuation of agricultural scientific and technological achievements, a method for dynamically evaluating the transaction price of agricultural scientific and technological achievements using computers was studied, providing data support and pricing basis for the transformation, transaction or transfer of agricultural scientific and technological achievements. [Method] A five-layer Fuzzy neural network model FNN (Fuzzy neural network) was constructed by combining fuzzy theory and neural network, which Abstract: [Objective] To solve the problem of difficult valuation of agricultural scientific and technological achievements, a method for dynamically evaluating the transaction price of agricultural scientific and technological achievements using computers was studied, providing dalearned and stored rule knowledge from the historical transactions data of agricultural scientific and technological achievements and used it to predict the transaction price of the achievements. The accuracy rate of FNN was only 80% after the actual business data test. In order to improve the evaluation accuracy, adapt to the changes caused by increasing transaction samples, reduce the resource consumption of retraining all sample data, so introduce incremental learning, an improved fuzzy neural network model IFNN (Improved fuzzy neural network) and an improved valuation process were proposed. [Result] In the actual business data processing, by applying the improved valuation process and completing the training of incremental data, the IFNN's valuation accuracy rate reached 86.7%. Fifteen rapeseed varieties were selected to compare the results of FNN and IFNN model. The fuzzy membership degrees of FNN and IFNN were different, and the comparison showed that the accuracy of IFNN was higher than FNN. The IFNN model algorithmically met the requirement that the estimation accuracy increased with incremental learning. However, in order to further improve the accuracy of the estimation, more complete training data was needed. [Conclusion] The IFNN can dynamically estimate agricultural scientific and technological achievements based on objective data in practical business, and can adapt to the constantly changing and developing society, and has high practical application value. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF