Evolving deep neural networks using coevolutionary algorithms with multi-population strategy

PM2.5 prediction based on modified whale optimization algorithm and support vector regression

In order to obtain the pattern of variation of PM2.5concentrations in the atmosphere in Nanchang City, we build a Support Vector Regression(SVR) with modified Whale Optimization Algorithm(WOA) hybrid model (namely mWOA-SVR model) that can predict the PM2.5concentration. Firstly, according to the Pearson correlation coefficient (PCC) method to examine the dynamic relationship between air pollutants and meteorological factors together with them, PM10, SO2and CO were selected as air pollutant concentration characteristics, while daily maximum and minimum temperatures, and wind power levels were selected as meteorological characteristics; then, using modified WOA algorithm for parameter selection of SVR model, four sets of better parameter combinations were found; finally, the mWOA-SVR model was built by the four sets parameters to predict PM2.5concentration. The results show that the prediction accuracy of mixed mWOA-SVR model with pollutant concentration plus weather factors as the feature was higher than single pollutant concentration.

A Survey on Evolutionary Neural Architecture Search

Deep neural networks (DNNs) have achieved great success in many applications. The architectures of DNNs play a crucial role in their performance, which is usually manually designed with rich expertise. However, such a design process is labor-intensive because of the trial-and-error process and also not easy to realize due to the rare expertise in practice. Neural architecture search (NAS) is a type of technology that can design the architectures automatically. Among different methods to realize NAS, the evolutionary computation (EC) methods have recently gained much attention and success. Unfortunately, there has not yet been a comprehensive summary of the EC-based NAS algorithms. This article reviews over 200 articles of most recent EC-based NAS methods in light of the core components, to systematically discuss their design principles and justifications on the design. Furthermore, current challenges and issues are also discussed to identify future research in this emerging field.

Correlation Analysis of China’s Foreign Trade Structure and Industrial Structure Based on Correlation and Mutual Influence

In recent years, China’s foreign trade has continued to develop and achieved remarkable results, and the economic development trend has been steady and positive overall. However, although there are new opportunities for the development of China’s foreign trade, it also faces new challenges and dilemmas. Therefore, in the current international trade situation, it is of great significance to study the impact of China’s foreign trade on industrial structure and find the key points for the development of endogenous economic dynamics on this basis. China’s industrial structure has undergone a clear process of upgrading and is a major trading country for industrially manufactured goods. The results of the analysis of the subsectors show that the trend of structural upgrading within the industry is equally significant. At the same time, there is often a significant trend of synergistic changes in the percentage of processing trade and the percentage change of trade in high-tech products. Based on this, this paper analyzes the correlation between China’s foreign trade structure and industrial structure based on the correlation of mutual influence. The evolutionary algorithm has the advantages of a strong ability to search for global optimal solutions and good robustness, which is applied as the core algorithm of this paper. The study in this paper examines the interaction between industrial structure and trade structure from the perspective of matching the association between industrial structure and trade structure on the basis of measuring the industrial structure and trade structure at the regional level. The experimental results demonstrate the validity of the model in this paper.

Evolving Deep Architecture Generation with Residual Connections for Image Classification Using Particle Swarm Optimization

Automated deep neural architecture generation has gained increasing attention. However, exiting studies either optimize important design choices, without taking advantage of modern strategies such as residual/dense connections, or they optimize residual/dense networks but reduce search space by eliminating fine-grained network setting choices. To address the aforementioned weaknesses, we propose a novel particle swarm optimization (PSO)-based deep architecture generation algorithm, to devise deep networks with residual connections, whilst performing a thorough search which optimizes important design choices. A PSO variant is proposed which incorporates a new encoding scheme and a new search mechanism guided by non-uniformly randomly selected neighboring and global promising solutions for the search of optimal architectures. Specifically, the proposed encoding scheme is able to describe convolutional neural network architecture configurations with residual connections. Evaluated using benchmark datasets, the proposed model outperforms existing state-of-the-art methods for architecture generation. Owing to the guidance of diverse non-uniformly selected neighboring promising solutions in combination with the swarm leader at fine-grained and global levels, the proposed model produces a rich assortment of residual architectures with great diversity. Our devised networks show better capabilities in tackling vanishing gradients with up to 4.34% improvement of mean accuracy in comparison with those of existing studies.