Image retrieval based on fireworks algorithm optimizing convolutional neural network

For the problem of the slow convergence rate of the traditional CNN algorithm using gradient descent algorithm to optimize parameters, a CNN model based on fireworks algorithm optimization is proposed, the parameters that need to be optimized are used as the input to the fireworks algorithm., and the cross-entropy (error) is used as the fitness function to improve the tuning process in the backward propagation., the MNIST handwritten character set is used for simulation experiments of image retrieval. The results show that the convergence speed of the improved algorithm is significantly improved.