Application of Neural Networks and Genetic Algorithms in Establishing Logical Rules for Evaluating the Edibility of Mushroom Data

The cultivation of mushrooms for commercial purposes has had a notable impact on the global economy, as evident from recent agricultural data. The mushroom industry is rapidly becoming one of the most lucrative industries in India and around the globe. This has led to a rise in popularity of mushroom farms and cultivation. The objective of this study is to determine the logical rules for edibility of mushrooms based on 22 biological characteristics. To achieve this goal, two different approaches were utilized. The first approach involved using Back Propagation Neural Networks that were constructed using data mining techniques. This method is distinct from a simple Decision Tree as it doesn’t require listing rules one by one. The cost function of the neural networks is based on the standard MSE (mean square error) plus a convergence controller. The Neural Network can be perceived as a tree-like logical graph, with a weight of 1 representing a strong positive correlation, −1 indicating a strong negative correlation, and 0 signifying no correlation. As a result, the top 4 logical rules account for every possible outcome based on most coverage. In the second approach, a Decision Tree is combined with a Genetic Algorithm (GA) for feature selection. The accuracy of the decision tree, which is constructed from the GA’s feature selections, feeds back into the GA’s fitness function. This method provides an alternative way of identifying edible mushrooms based on their biological characteristics.