Online extreme learning machine target identification method and system based on consistent regularization

Abstract

The invention discloses an online extreme learning machine target recognition method and system based on consistent regularization, and the method comprises the following steps for carrying out the learning and training of a classifier based on a single implicit strata feedforward neural network: carrying out the feature extraction of a training image, and obtaining a corresponding image feature set, randomly dividing the image feature set into a plurality of feature subsets; respectively generating corresponding neighbor feature samples for the divided feature subsets; introducing consistent regularization constraint into an online extreme learning target optimization function, randomly generating hidden layer node parameters of a single hidden layer feedforward neural network, and selecting any feature subset and adjacent samples thereof to perform initialization network weight generation; and performing iterative updating of network weight based on the residual feature subsets to complete learning and training of the classifier based on the single hidden layer feedforward neural network. The method has the advantages of being high in noise tolerance, high in classification and recognition precision, high in learning speed and good in task expandability.

Description

technical field [0001] The invention relates to the technical fields of image classification and target recognition, in particular to an online extreme learning machine target recognition method and system based on consistent regularization. Background technique [0002] Extreme learning machine (extreme learning machine, ELM) is an efficient, generalized neural network learning algorithm based on a single hidden layer feedforward neural network. The theory of extreme learning machine is inspired by biological learning and tries to explain and answer the basic question of whether biological neurons need to be adjusted during the learning process. Shortly after it was proposed, Harvard University, Stanford University, MIT University, IBM Watson and other well-known research institutions successively obtained direct or indirect verification in the olfactory system of mice, the visual system of monkeys and the perception system of humans. The core content of the theory of extr...

AI Technical Summary

Problems solved by technology

However, although the classifier trained by the extreme learning machine based on deep network features has significantly improved classification performance and efficiency, it will be affected by the noise existing in the input and its corresponding label, and lead to inevitable performance degradation

The reason for this problem often lies in the inherent shortcomings of traditional extreme learning machines, that is, traditional extreme learning machines are still based on empirical risk minimization, and overfitting is likely to occur under noise interference.

Therefore, the effect performance of target recognition technology based on traditional extreme learning machines is far from satisfactory, and often needs to be improved by introducing certain regularization constraints.

Images