Back propagation method for out-of-order data stream in big data

Abstract

The invention provides a back propagation method for an out-of-order data stream in big data. An improved back propagation algorithm based on dynamical adjustment (IBPDA) is provided to solve the problem that association rules are difficult to obtain from the out-of-order data stream in the big data. A dynamic self-adaptation structure adjustment mechanism is used, a network training structure is adjusted in a self-adaptation mode according to environment requirements, invalid training nodes are automatically deleted, and optimized iteration of the training process is achieved; three factors of a neural network, namely a learning index, a momentum factor and a scale factor, are dynamically adjusted in the web-based learning process to achieve the aims of increasing learning response speed and enhancing network stability. As is shown in a simulation result, by means of the dynamic self-adaptation structure adjustment mechanism and dynamic adjustment of the three factors of the neural network, the method can obtain more convergence times, effectively improve the convergence rate and improve whole network performance.

Description

technical field [0001] The invention is an improved backpropagation method for out-of-order data streams, belonging to the field of data processing in big data. Background technique [0002] Big data, or huge amount of data, involves a huge amount of data, and it is impossible to obtain data association rules in disordered data within a reasonable time through current mainstream software tools. The traditional data processing mode is that humans are active and data is passive. The collected data is first stored in the database management system, and then the user actively queries to get the final answer. However, this method is not suitable for massive and endless real-time data streams. not suitable. The backpropagation algorithm (BackPropagation, BP), referred to as the BP algorithm, is an effective learning prediction algorithm that can perform large-scale parallel information processing, has a strong simulation ability for nonlinear systems, and can effectively predict ...

AI Technical Summary

Problems solved by technology

However, the traditional BP algorithm still has some shortcomings. In specific applications, the learning rate of the network is often fixed. When the number of iterations increases, the performance of the network will deteriorate, which makes the learning speed of the algorithm and There is a big contradiction between the stability of the network

In addition, there are inherent defects in the BP algorithm, such as: the network structure is difficult to determine, the hidden layer structure is often determined through experience, the training time is long, the convergence speed is slow, it is easy to fall into a local minimum, and the prediction effect is not good, etc.

[0003] The traditional BP algorithm mainly includes the following problems: (1) The adjustment rate is relatively slow due to the increase in the number of iterations

(2) It is easy to fall into a local minimum during the learning process

(3) Defects of convergence rate and convergence accuracy

(4) Selection of structured parameters and learning rate in artificial neural network

(5) The BP algorithm usually applies the Sigmoid function, and when the Sigmoid function enters the saturation region, the weight correction amount in the weight correction formula becomes negligible, which makes the network training fall into a saturated state, greatly reducing the learning efficiency

Images