Research On Deep Learning Design In Smart Classroom

The global movements of 21st-century skills(key competencies)and the reemergence of artificial intelligence have made deep learning re-emerged in the academic community and the masses.In the field of educational technology,it has become the main goal and appeal of IT-based teaching to move towards deep learning.In this regard,this dissertation conducts in-depth research on the research trends and state of the deep learning and the evolution of its concept through literature analysis.Based on the results,the concept and characteristics of deep learning are also newly defined.Besides,the results of the literature analysis also show that there is no research focuses on how to promote student deep learning in a flexible way which in the context of the smart classroom environment by analyzing the relevant literature at home and abroad.What matters is that this flexibility is exactly the demand for deep learning.To fill this gap,this dissertation deeply analyzes the necessity of deep learning from the pedagogical structure that characterizes pedagogical stability to the learning architecture that characterizes pedagogical flexibility(not to negate the pedagogical structure,but to focus on the other side of pedagogy),and the potential of the smart classroom to support flexible deep learning.And based on these,this dissertation attempts to deal with the following four issues: a)What does a deep learning architecture for a smart classroom look like? b)How to instruct teachers to do deep learning design for smart classrooms? c)How to provide a learning scaffold for students' deep learning? d)What are the practical effects of this deep learning in the context of smart classrooms?The issue a)focuses on guiding ideas in the top-level,which aims to define the deep learning architecture and build its model.To this end,the definition of deep learning architecture is defined through two-time inquiry analysis which begins with the word meaning of "architecture" and the definition of it in both fields of architecture and computer science.And then,the deep learning architecture model for the smart classroom is constructed which can reflect the flexibilities between effectiveness and interesting of learning tasks,between instruction and self-direction of the learning activities,between order and disorder of learning process,and between teacher' datainformed and machine's data-driven of teaching decision-making,which are based on the visual perception of the existing typical learning architecture model and combined with the functional characteristics of the smart classroom environment.Issue b)focuses on design approaches in the middle-level,which aims to develop a deep learning design framework that fits the smart classroom with the flexibility concept of deep learning architecture.To this end,this dissertation clarifies how this framework achieves the above four flexibilities of deep learning through instructional design and the behaviors of smart classrooms,teachers,and students based on the concept of deep learning architecture made in the issue a)and the processes of literature search,construction of the prototype,validation of the prototype,and finalization of the framework.This dissertation also clarifies how each part of this framework should be designed to promote the students to learn flexibly and deeply supported on the smart classroom through the above processes.The issue c)focuses on supporting tools in the bottom-level and aims to design a kind of deep learning scaffold catering to the concept of deep learning design framework.To this end,a deep learning sheet severing for students' deep learning in a smart classroom is developed based on the above deep learning design framework developed in issue b)and stages of prototype construction,teachers' trial feedback,iterative correction,and quality assessment.And further,the structure of deep learning objectives is simplified,the general design process of evaluation evidence supported by intelligent equipment and technology is added,and the backward design sequence of learning tasks,the task presentation template conducive to the flexibility of deep learning,and the task representation strategies and principles supported by rich media are supplemented.Issue d)focuses on the practical effects of deep learning and is designed to assess the actual practicality and effectiveness of deep learning sheets.To this end,one excellent teacher was selected by purposive sampling to conduct a 6-week educational experiment by using a new kind of interactive flexible teaching modle of deep learning in the two classes(infused smart classroom environment)he taught.This teacher had participated in the trial and evaluation of the deep learning sheet.The educational experiment examined the practical effects of deep learning in the four aspects of deep engagement,deep learning strategies,high-level knowledge development,and transfer.The data analysis found that the deep learning sheet of infused smart classroom can improve the engagement of deep learning,can guide students to use deep learning strategies,and can promote the development and transfer of high-level knowledge,but this only suits for students with the highly creative experience.At last,it can be concluded that though there are some blemish and inadequacy,the proposed deep learning architecture is reasonable,the backward design framework of deep learning is effective,and the deep learning sheet is useful after the in-depth analysis and discussion.