Yixue Squirrel AI Learning Appears at ECML-Pkdd 2018 in Dublin, Talking About New Developments in AI-Enabled Education
The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery 2018 (ECML-PKDD) was successfully held at Croke Park Conference Center in Dublin, Ireland. More than 600 well-known professors and scholars from universities around the world, as well as entrepreneurs and experts in the field of computer science, had in-depth discussions and exchanges on computer science, machine learning, data mining, knowledge mining and other related topics and shared the latest cutting-edge research findings. Squirrel AI Learning’s Chief Scientist Dr. Wei Cui attended the conference on invitation. He gave an insightful speech and shared the stage with many famous professors and scholars, such as Naftali Tishby, a physicist/neurologist and a professor of computer science from the Hebrew University of Jerusalem, Corinna Cortes, a Danish computer scientist and the person in charge of New York’s Google Research, and Martin Wistuba, a data scientist of IBM. Dr. Wei Cui presented the technological advantages of Squirrel AI intelligent adaptive system and the achievements in technology-driven application. He also shared the development and application of AI in education and how it… became the area of focus for researchers and companies from all over the world.
It is widely believed that ECML-PKDD is a top conference on machine learning and data mining in Europe. It has been successfully held in various places in Europe for more than 16 consecutive years. The topics of the conference mainly include machine learning, AI, data mining and their applications, covering social, economic and other important fields today. It’s an international event with far-reaching influence in the global science and technology sector.
Squirrel AI Learning’s Chief Scientist, Dr. Wei Cui gave a wonderful keynote speech at the conference. According to Dr. Wei Cui, Squirrel AI intelligent adaptive system is a “complete cycle” AI-powered product launched by Yixue Education, which completes the core process of teaching and learning under the guidance of the system. It applies the intelligent assessment algorithm, ability diagnosis and student status representation model, as well as the recommendation algorithm applied in learning path and learning content planning. In addition, Dr. Wei Cui revealed that Yixue is developing algorithms for learning mode selection and early intervention using deep learning. The core is to collect and analyze learning data for AI to use the “nanoscale” knowledge graph to assess students’ target knowledge mastery in the shortest time. Therefore, it establishes personalized dynamic student portraits by measuring students’ knowledge status constantly, to understand each student’s study status and problems, by which it designs testing and learning paths accordingly, and adjusts the teaching behaviors. Meanwhile, it constantly recommends the most appropriate learning materials in the process of learning, measures the learning effects, and establishes self-learning models and provides feedback on AI prediction and the effects of the content.
Dr. Wei Cui introduced three modeling elements of Squirrel AI, which respectively are: (1) user portraits for students, i.e. the mastery of each student’s personal preference and interests, learning style, cognitive characteristics, ability level and state of knowledge; (2) learning content modeling and “nanoscale” knowledge graph building, i.e. to present different forms of learning resources in the form of videos, texts, audios, pictures and exercises, establish an algorithm to label knowledge points and exercises, and give the corresponding difficulty coefficients, etc.; (3) personalized matching, i.e. to match the most suitable learning paths and courses and recommend personalized learning content for each student based on the data generated in the first two steps, to maximize learning efficiency.
At present, apart from the chain of nanoscale knowledge point decomposition, correlation matrix and knowledge graph, Squirrel AI also recommends personalized learning paths, which are all at globally leading levels. Of course, that is not to say that robots teach students like teachers. Just like Alpha Go judges the situation of a chess game by memorizing the chess manual + machine learning, Squirrel AI precisely identifies every student’s existing knowledge points through knowledge graph and the information theory. In the past, there was no clear user portrait for each student. We could only tell whether a student was a 50-point student or a 95-point student, but couldn’t tell the student’s actual knowledge mastery based on his knowledge graph. Now, algorithms and techniques can solve this problem. Through strategic logic deduction, Squirrel AI intelligent adaptive system gives students different teaching videos and solutions to different exercises, including videos, animations, recordings, lecture notes and various exercises.
That is to teach students in accordance of their aptitudes, which we are always stressing in education. At present, there is still no way to teach students according to their aptitudes in the classroom. All students listen to the teacher at a fixed pace, whether you are a 50-point student or a 95-point student. Such uniform education is like Tencent news and Sina news in the past. If we want to distribute different contents to different users like Toutiao does, we need machine learning and genetic algorithms, to let the system constantly and more and more accurately push exercises and contents that match each student’s weak points, to achieve the most efficient learning effects.
At last, Dr. Wei Cui mentioned a recent human vs. machine teaching competition organized by Squirrel AI Learning in 100 cities across China. Squirrel AI intelligent adaptive system competed fair and square with experienced public-school teachers. The results showed that the AI adaptive learning system is superior to human teachers in teaching quality and effects. In addition, Yixue Education has established a joint lab for AI adaptive learning with Stanford Research Institute (SRI) and a joint lab for parallel AI intelligent adaptation with CAS, to jointly explore the continuous technological innovation capability in the era of AI in the future.
The ultimate dream of Squirrel AI Learning is to let every child have an AI teacher like a combination of Socrates + Da Vinci + Einstein, to teach them. In the future, no school district office will be needed in the world. Great mentors will be available not only for princes. We can use an AI system to provide a highly intelligent teacher to any child, to bring out their potential and achieve true educational equality.