Deep Learning Methods And Their Intelligent Application Considering Uncertain Information And Differential Structure

Due to the deepening dependence and integration of the global economy,the COVID-19 pandemic has hindered the world industrial chain and supply chain cycle,increasing the vulnerability of the global economy.However,while the COVID-19 epidemic has had a great impact on global economic and social development,it has also brought demand stimulus to emerging technologies such as big data,artificial intelligence,and cloud computing.These technologies can not only have an important impact on epidemic prevention and control,social governance,and resumption of work and production but also play a key role in the global economic recovery and even further development in the post-epidemic era.Based on this,as the driving force for the development of a new round of intelligent technology,the importance of the above emerging technologies has gradually emerged.Meanwhile,it also increases people’s demand for the development of intelligent technologies in actual production and life,which also brings opportunities for the further development of intelligent technologies.For countries or enterprises,there are still some issues that need further consideration in the intelligent technology market,such as intelligent technology evaluation and decision-making.However,as a rapidly developing emerging thing,it is often difficult for intelligent technology to obtain sufficient quantity or effective quantitative data in its evaluation and decision-making process,then qualitative data shows its necessity and importance.In the actual evaluation and decision-making process,qualitative data can usually be described by a variety of uncertain information expression methods.As for the possible hesitation and fuzziness of uncertainty evaluation,hesitant fuzzy information can completely and reasonably describe it.Based on this,the uncertainty information involved in this paper is hesitant fuzzy information,and this information can be further introduced as the research basis of intelligent technology evaluation.In addition,hesitant fuzzy information is mainly used in the general decision-making field,but less in the evaluation or decision-making of intelligent technology selection.Therefore,this paper further designs the deep learning method considering the differential structure and fuses them with various hesitant fuzzy information adapted to different data characteristics to apply to appropriate intelligent technology evaluation.For example,the discrete,parallel,seriesconnection,and continuous information are respectively fused into the convolutional neural network(CNN)model,the generative adversarial network(GAN)model,the deep belief network(DBN)model,and the recurrent neural network(RNN)model,to propose a deep learning method considering differential structure.These deep learning methods in the hesitant fuzzy environment are applied to different types of intelligent technology selection processes.Therefore,combining multi-form hesitant fuzzy information,differential structured deep learning methods,and different types of intelligent technology selection can effectively deal with the intelligent technology selection issues where quantitative data is unavailable or difficult to obtain in a relatively short time.According to this,the main research contents of this paper can be presented in the following contents:This paper designs the deep learning method considering uncertain information and discrete CNN structure and applies it to the image and text recognition technology,proposes the deep learning method considering uncertain information and continuous GAN structure and applies it to the intelligent monitoring technology,develops the deep learning method considering uncertain information and parallel DBN structure and applies it to the news push technology,and constructs the deep learning method considering uncertain information and series-connection RNN structure and applies it to the automatic driving technology.According to the proposed deep learning method considering uncertain information and differential structure,this paper can improve the discrete,continuous,parallel,and seriesconnection information representation methods,and provide a reference for further research of intelligent technology selection using multiple hesitant fuzzy information.Based on this new hesitant fuzzy information,this paper can design deep learning methods considering differential structures,namely,deep learning methods considering discrete CNN structure,continuous GAN structure,parallel DBN structure,and series-connection RNN structure,which provides the theoretical basis for further research of the deep learning methods considering the differential structure.According to the above multiple forms of multi-form hesitant information and the deep learning method considering the differential structure,this paper can optimize the decision-making tools for the image and text recognition technology,the news push technology,the automatic driving technology,and the intelligent monitoring technology selection,providing theoretical support and application basis for the further selection of different intelligent technologies.The above content can realize the optimization of the model architecture and the application of the intelligent technology field,thus it is feasible and reasonable to apply deep learning methods considering uncertain information and differential structure to different types of intelligent technology selection.