Work

Multiple Kernel Learning for Learner Classification

Year: 2023
Type: conference-paper
Abstract: :At present, educational researchers are continuously exploring new ways of learner models. To solve the problems of data bias and feature mismatch that may occur in learner modeling on heterogeneous data, this paper uses the method of Multiple Kernel Learning (MKL) to accurately portray the features of learners' thinking skills and construct a classification model for top talents. First, by analyzing the thinking skills that have the greatest impact on top talents, the features of critical thinking skills, logical thinking skills and computational thinking skills are extracted separately. Then, a learner classification model for top talents was constructed by MKL method. Finally, the model was validated on the dataset and compared with the results of single-kernel learning. The experimental results show that the accuracy of the classification model for top talents based on MKL is significantly better than that of ordinary single-kernel learning, which indicates that the method of MKL performs better in the learner classification model.
Language: English
Open Access status: bronze
OpenAlex ID: W4391893128
FWCI: 0.357
Cites: 11
Cited by: 2
Related to: 10