Multiple Lesions Detection of Fundus Images Based on Convolution Neural Network Algorithm With Improved SFLA

Abstract

In order to effectively solve the problem of interlaced overlap in the fundus image lesions, large and small blood vessels packed densely and severely affected by light, and to achieve multi-label classification of fundus images. In this paper, a single population leapfrog optimization convolutional neural network algorithm (SFCNN) is proposed to detect and classify various fundus lesions. The algorithm uses the efficient search ability of the shuffled frog leaping algorithm to optimize the weight initialization and back propagation of the convolutional neural network. In order to deal with the problem of fundus image classification in the big data environment, the novel grouping optimization strategy is presented to effectively combine Spark platform and SFCNN algorithm to achieve large-scale fundus image classification and detection of multiple lesions. The experiment of the detection of fundus image lesions shows that the accuracy rate of SFCNN is better improved in both single lesion detection and overall detection, compared with other algorithms.INDEX TERMS Shuffled frog leaping algorithm, convolutional neural networks, fundus image, detection of multiple lesions, weight optimization.

Deep ConvNet: Non-Random Weight Initialization for Repeatable Determinism, examined with FSGM

“…With the existing random number methods several learning sessions are required to establish which learning session's random sequence has provided the best accuracy from a variation of random initialization states. In comparison to Blumenfeld et al [12], Ding et al [13], Wang et al [14] and Ferreira et al [15] approaches, that required to adjust weights, use random numbers or sampling the dataset for convergence, the proposed method, is without data sampling or random sequences as a more general case, and is a complimentary approach to both Glorot/Xavier, and He et al initialization limit values [16]. The proposed method substitutes only the use of random numbers for a deterministic non-random finite number sequence, and retains the number range limits of the original methods [16].…”

“…Ding et al in 2020 [13], proposed a shuffle leap frog algorithm approach, for the update and initialization with random Gaussian forms in the area of fundus lesions images. The approach presented in that paper contains random numbers, initially in a Gaussian distribution optimised with the shuffle leap frog algorithm, where as the approach presented in this paper, does not contain random numbers.…”

“…However, if there is a proceeding convolutional layer, then the activations map to the receptive fields of the convolved image filters, which is the case in this benchmark model. So in this case the height and width are the image size and the depth is inherited as the channel depth (or filters from the previous layer), as in Equation ( 12), and the value set of cnt is in Equation (13).…”

“…The FSGM transferred learning approach used is by Theiler [26], and it explains the attack with examples, and importantly uses the same MNIST dataset. See Figure (13) for the Theiler and Torres architecture as integrated, with three experiment test points. [26], as added to the Torres's benchmark model [18].…”

Study on Optimal Control Strategy for Cooling, Heating and Power (CCHP) System

“…(1) Cauchy mutation operator Cauchy distribution is a kind of functional distribution commonly used in mathematical statistics and other fields. Its probability density distribution function can be written as follows [28][29][30].  == is satisfied, it is called the standard cauchy distribution, denoted by C(0,1).…”