Deep matrix factorization based on self-attention mechanism for student grade prediction
| Autor: | Guo Zongxin, Liu Tieyuan, Gu Tianlong, Chang Liang |
|---|---|
| Rok vydání: | 2020 |
| Předmět: | |
| Zdroj: | Journal of Physics: Conference Series. 1651:012019 |
| ISSN: | 1742-6596 1742-6588 |
| Popis: | Deep learning is gradually emerging in the field of educational data mining. Aiming at student grade prediction in education data mining, a prediction model combining self-attention mechanism and deep matrix factorization (ADMFN) is proposed. It can quickly extract important features of sparse data and process complex nonlinear data. Firstly, the student and course vector corresponding to the score matrix are input into the model and projected to obtain the potential feature vector of the student (course). Then, the potential feature vectors of students (courses) are added into the self-attention mechanism to construct a multi-layer perceptron network. Finally, by constructing a bilinear pooling layer, the output vectors of rows and columns are fused to obtain the predicted value of the score. We compare the existing recommendation system baselines to evaluate our model. The experimental results show that the proposed model is effective in real dataset. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|