Machine learning and human learning - what can one learn from the other?
Invited conference paper presented and published in conference proceedings



摘要Human learning, or education, is a complex process that depends not only on the individual student’s unique personalities, capabilities, prior knowledge and believes, but also on his/her interactions with the teacher, the peers, the learning environment, and lastly, the content knowledge. Traditionally this process is studied in the field of education and psychology. Due to the complexity of these human nature and environmental influences, there is a huge potential to apply machine learning techniques to study and understand the process of education. We have adopted machine learning techniques in developing our educational model on data collected from various sources of a foundation course in the Chinese University of Hong Kong. This model includes objective data such as student grades, their prior knowledge, their language capability, the learning environmental conditions, and qualitative or subjective data such as essays and questionnaires that directly assess students’ reflections and perceptions on their learning process and outcome. This mix of data is not commonly found in other applications of machine learning, yet provided a platform for it to excel. Preliminary result of our study indicates that our model has achieved an improvement in its predictive power on student’s final achievement. The study provides insights on how machine learning can be used to help us to understand and improve our own learning process. This presentation will also shed some lights on what the machine learning community could gain from educational theories.
著者Kiang Kai Ming
會議名稱International Conference on Deep Learning Technologies 2018
關鍵詞Machine Learning, General Education

上次更新時間 2018-08-11 於 12:28