Research On Scientific Philosophy Of Deep Learning

As one of the key research areas of artificial intelligence,Deep Learning is a hot topic in the disciplines of computer and software engineering in recent years.It intersects with disciplines such as brain science,mathematics and physiology,and promotes communication and interaction between various disciplines.The rapid development of Deep Learning is inseparable from the huge amount and fast processing of big data science and technology,and has been applied in the fields of natural language processing,image recognition,and unmanned driving.The so-called Deep Learning is to classify and visualize data through artificial neural networks to achieve unsupervised learning.The essence of Deep Learning is based on the brain.In the process of constructing and simulating the human brain,it is based on the input and output of the data,and allows the data to achieve low-level to high-level feature extraction,and the computer progresses at the data layer.In the process,computer gradually "understand" and "learn" the data and obtain the corresponding knowledge information,that is,Deep Learning is a process of forming a more abstract high-level representation by combining low-level features,and then achieving the goal of optimal feature learning.Through the analysis of domestic and foreign literature,it can be found that foreign philosophical research on Deep Learning can be traced back to the 1950 s of artificial intelligence philosophy.Deep Learning is a further development of machine learning;domestic research on Deep Learning has expanded from the computer field itself to the philosophical society in the field of science,it mainly focuses on the philosophical research of artificial intelligence,the comparative analysis of Deep Learning and human intelligence,etc.The research perspective includes epistemology and methodology,etc.The research on these aspects is still in the ascendant.Therefore,this article attempts to discuss Deep Learning from the perspective of philosophy of science,from three aspects of epistemology,scientific methodology and ethics.This article sorts out the concepts and periodical evolution of learning and Deep Learning.On the basis of dividing Deep Learning models into Supervised Deep Learning models,Unsupervised Deep Learning models and Mixed Deep Learning models,the following perspectives are carried out.The investigation:(1)The perspective of epistemology.Based on Marxist epistemology,the cognitive subject,cognitive object and cognitive tools of Deep Learning are analyzed,and it is believed that although Deep Learning expands the scope and number of cognitive objects and makes the cognitive objects appear in the form of virtualization,it cannot Making machines replace humans as the new subject of recognition,but only enlarges and extends the cognitive function of humans,and becomes an important tool for humans to understand the world.Deep Learning makes "data" the main source of knowledge and improves the status of verbal knowledge,but at present it seems that computer "data" and data processing are very different from human conscious activities.(2)Discussion on the scientific methodology of " Deep Learning ".It is believed that Deep Learning has brought new content and special expressions to scientific methods such as inductive reasoning,deductive reasoning,and system analysis.(3)Ethical reflection on deep learning.The problems of privacy invasion,data alienation,employment safety issues and physiological ethics that may be brought about by Deep Learning and big data processing are combed,inspected and analyzed,and it is proposed that "peopleoriented" can effectively deal with these problems.