TORONTO, June 26, 2019 /PRNewswire/ -- Recently, the AERA 2019 was held under the title of "Integrated Application of Technologies to Mathematics Teaching and Learning" in Toronto, Canada. Sam Wang, a senior data scientist at the SRI International, was invited to the summit and gave a speech titled "Putting Technology to the Test: Efficacy Studies of an Adaptive System in China" on behalf of Squirrel AI Learning by Yixue Group, describing to the other participants a wonderful blueprint of fully utilizing the Internet, AI technology and high-level educational resources to enhance the personalization of educational content.
Sam Wang, a senior data scientist at the SRI International, is making a presentation on behalf of Squirrel AI Learning
The host American Educational Research Association (AERA) is one of the largest U.S. education industry associations, a globally influential educational academic organization, and the largest national inter-disciplinary research association. Founded in 1916, the association is committed to upgrading and disseminating educational knowledge, encouraging scholars to do research into education and promoting the improvement of educational and public services through scientific research.
The AERA Annual Meeting is the world's largest gathering of education researchers and is attended by more than 14,000 AERA members, as well as scholars, policy experts and practitioners from related fields of discipline. The AERA Annual Meeting is always focused on the showcase of groundbreaking research and innovations, including studies of all the educational phases from early education to higher education.
Felice J. Levine, the executive director of AERA, talked about the AERA 2019 and said, "The AERA Annual Meeting is an excellent opportunity for educational researchers to share important new discoveries and innovations. These studies and innovations will lead to a continuous optimization and improvement in educational practices and policies.
It is worth noting that a number of speeches on digital technology and higher-order analysis were given at the AREA 2019. The speech titled "A Bayesian Look at Reliability", given by Charles Lewis, a professor at Department of Psychology, Fordham University, was one of the eight keynote speeches made at the annual meeting. In the speech, Charles Lewis discussed the challenge posed to the reliability of the Bayesian classification algorithm when it was used for modeling in the education industry. The research project led by Charles Lewis won the E. F. Lindquist Award at the AREA 2018.
Dr. Lewis expounded the principles of test reliability in simple terms in his speech. He also explained their similarities and differences in the CTT and IRT models from the traditional perspective and Bayesian perspective. Then, he further pointed out, "Either of the pedagogical testing and measurement theory is better or worse than the other from the traditional perspective and Bayesian perspective, but we can deepen our research of this issue on a larger scale from different perspectives."
As a pioneer in China's education industry, Squirrel AI Learning attended this grand event of education and gave a speech titled Putting Technology to the Test: Efficacy Studies of an Adaptive System in China, drawing high attention from the attendees. Dr. Sam Wang showed the rest the difference in teaching effect between the Squirrel AI Learning system and human teachers in different teaching scenes and class sizes. The SRI International found in their research that after using the Squirrel AI Learning system, the students in both big classes (20-30 students) and small classes (about 3 students) improved their academic performance more significantly than the students in the control group.
As we all know, the Internet has produced a subversive impact on all traditional industries. Like all the other traditional industries, the education industry has begun to introduce digital means to improve the educational effectiveness and educational resource utilization rate.
Subversion usually begins with a response to challenges hard to overcome by the traditional industries. As we know, teaching and learning are two sides of an educational activity: there should be not only high-quality content, but also an individualized learning plan. However, the traditional education industry is just facing the following two challenges: prevailing disequilibrium of educational resources and lack of individualized educational content.
The revolution in education on the Internet aims to eliminate the disequilibrium of educational resources first.
Geographical limitations and different levels of economic development contribute to the difference in teaching strength from region to region, and many students have no way to acquire high-quality local education resources. Fortunately, "online schools" have risen at the historic moment. Academy, Coursera and New-oriental Teach-on-line, have sprung up quickly and achieved fast expansion.
However, the MooC as an "online school" has obvious defects. On the one hand, due to low entry barriers, lack of classroom discipline management, and shortage of support from a learning community, learners feel it easy to study online at the beginning, but they also feel it "easy" to give up learning. On the other hand, due to unified content of courseware, personalized content or learning plans cannot be provided for students that have different knowledge structures and learning capacities. In the end, the two reasons lead to a low course completion rate of "online education". A large number of learners give up halfway due to their poor self-discipline or failure to adapt to the teaching content.
As there is increasing demand for personalized education, "one-to-one" long-distance education has come into being on the Internet to solve the contradiction between the demand for personalized education and the lack of personalized education.
When one teacher teaches one student "face to face" on the Internet, specific educational services can indeed be offered to improve the quality. The teacher can adjust the teaching content in a timely manner according to the student's mastery of knowledge, and thus provides intensive training for the student. But this model also has very obvious weaknesses. The first is the teacher's low teaching ability. In most cases, "one-to-one" teaching cannot be truly provided in a personalized way, because there are, after all, only a limited number of high-level teachers, while most teachers just have average teaching abilities. Second, "one-to-one" teaching is costly, and one teacher can merely offer a service to one teacher each time.
So in the final analysis, neither "online school" education nor "one-to-one" teaching can ameliorate the above two issues, i.e., the disequilibrium of educational resources and lack of individualized educational content. These two challenges can only be radically resolved with artificial intelligence (AI), e.g., the machine learning model can be used to meet individualized needs while raising teaching quality to a high level.
As AI technology has become increasingly mature in recent years, many industries have begun to enable the use of AI. Educationally developed countries such as the USA have achieved many positive results in using AI for personalized teaching. But Asia, including China, hasn't widely applied AI to the education industry yet.
Rome was not built in a day. Adaptive education, which originated in the USA, has accurately located the bloodline of the educational reform. However, due to the imperfection of system functions and the high requirements for education providers' assessment and teaching and research capacities, the adaptive education always developed slowly in the past. However, on the eve of the AI revolution, the empowerment from AI brightened the future of the adaptive education.
The AI technology-based adaptive learning system can not only further intensify stratification, but also clearly identify any imperceptible change between each two capacity levels. According to students' real-time learning status, the system can dynamically adjust the next step of learning content and path, and build a more complex learning model to provide personalized education on a large scale.
The world's first AI adaptive education provider Knewton has offered services to 20 million North American students. RealizeIt, the largest provider of personalized and adaptive learning products in the U.S. higher education sector, offers more than 40,000 adaptive courses.
As the first Chinese education provider to use AI for education, Squirrel AI Learning has achieved remarkable achievements in primary and secondary education.
Squirrel AI Learning primarily provides AI adaptive learning programs services for primary and secondary school students. Squirrel AI Learning has successfully developed an advanced algorithm-based intelligent adaptive learning engine with complete independent intellectual property rights. The Squirrel AI Learning system can work as an expert teacher, but its efficiency is 5-10 times as high as the traditional educational efficiency.
The Squirrel AI Learning system has four main characteristics:
The core teaching content is taught, recorded and analyzed on the Internet, while a good learning environment is created at the offline training center to enhance the concentration on learning and ensure a reasonable time schedule.
The AI intelligent learning system teaches students in accordance with their aptitude, remedies weaknesses, timely identifies and helps master fail-to-master knowledge points, and assists in building a complete knowledge structure in which different knowledge points are connected to one another. The human instructor creates a group-learning atmosphere, adjusts students' attitude towards learning, and guides and encourages them to improve learning the efficiency.
The panoramic knowledge map helps students master knowledge points and clearly understand the relationship among different knowledge points. The multi-level knowledge points can be learned little by little and finally summarized by students in a scientific and effective way.
Students can clearly see their current knowledge structure graph with their knowledge application ability tested. Then the shortest path can be designed using the analytical model built based on Bayesian Discrimination for an improvement in students' knowledge level. Moreover, students' learning path can be adjusted at any time based on feedback from them so that they can maximize the learning efficiency.
Years of educational practice has confirmed that small class teaching is of great help to improve the quality of education. In order to verify the educational effectiveness of the Squirrel AI Learning system, Squirrel AI Learning organized teaching comparison testing to confirm whether:
Squirrel AI Learning did a control experiment by choosing 163 students in Grade 8 at a middle school in Sichuan Province to answer the first question. The students were aged 14-15. The experiment implementer SRI randomly divided these students into experimental group and control group, and let them taught by the Squirrel AI Learning system and outstanding teachers respectively. Each of the students in the experimental group was equipped with a dedicated computer. The control group was taught by mathematics teachers that had won an award in Sichuan Province, with 20-30 students in each class.
Before the experimental group and the control group formally started learning, the SRI International organized an examination on the knowledge points to be learned in order to identify each group's knowledge level.
After the test ended, the control group and the experimental group studied for 5 hours and 50 minutes in the following three days under the same conditions, including learning duration, rest duration and class activities schedule. After the end of learning, the SRI International tested the students' mastery of knowledge points again. The test paper was designed and reviewed by outstanding teachers who were not involved in the testing. 20 core contents totaled 100 points.
In order to avoid interference from other factors, the SRI International also collected basic information including the students' age and gender and their parents' education level.
The test results show that for question 1, i.e., standard whole class teaching, the grades in the experimental group increased by 1.58 points averagely, while that in the control group increased by 9.11 points on average. An analysis of covariance confirms that there is a statistically significant difference between the two groups.
For question 2, the SRI International adopted an almost same experimental method and selected some students from Shandong Province and divided them into the experimental group and control group. The students in either group studied for 8 hours and 30 minutes. There were 3 students in each small class.
The experimental results show that the grades in the experimental group increased by 6.96 points averagely, while that in the control group increased by 2.45 points on average. An analysis of covariance confirms that there is a statistically significant difference between the two groups.
The above control experiments clearly show that the Squirrel AI Learning system has obvious advantages for both the traditional class teaching and the small class teaching that boasts high learning efficiency. One of the advantages is that grades can be increased significantly, and the other is that it doesn't matter whether the teacher's teaching level is high or low. With the aid of the Squirrel AI Learning system, the government and educational institutions can quickly and inexpensively promote the best educational resources to any part of the country that can access the Internet. This will greatly promote educational fairness, improve the educational efficiency and increase the service efficiency of educational resources, and lay a solid foundation for the AI technology in comprehensively reforming the education industry.
Squirrel AI Learning founded a Yixue AI Lab with SRI as the primary research partner two years ago, and the AI laboratory is committed to carrying out the work based on the SRI International's unique advantages in AI and educational technology. Currently, the lab is making a splash in three key collaborative areas: 1) core adaptive education model and technology; 2) natural language processing and semantic analysis, aimed at realizing the virtual personalized assistance (VPA) function and using dialogue-based interface to diagnose students' errors and receive feedback on AI from instructors and students. 3) The multimodal integrated behavioral analysis (MIBA) research enables Squirrel AI Learning to understand students' emotional and psychological states and better predict their behavior and in a adaptive learning environment, and provide signals for human instructors to take intervention, remedy and support measures or remind, recommend, and relax students through the system.
Squirrel AI Learning is a leading Chinese AI education application provider, and is active in attending global education conferences. At the AERA Summit, it briefed North American education experts on the fast development and latest research results achieved by the Chinese education industry, and conducted in-depth exchange of views with the North American education experts present, especially the experts in data science and AI application.
We believe that with the help of the AERA, we will effectively promote the development of the Chinese education industry and the implementation of advanced educational technologies in China. Driven by advanced AI enterprises including Squirrel AI Learning by Yixue Group, China's education industry will soon fly up on the wings of AI.
View original content to download multimedia:http://www.prnewswire.com/news-releases/squirrel-ai-learning-by-yixue-group-gives-a-thesis-presentation-at-the-aera-education-summit-on-innovative-educational-and-learning-styles-300875192.html
