Integrated Sparse Coding With Graph Learning for Robust Data Representation

Abstract

Sparse coding is a popular technique for achieving compact data representation and has been used in many applications. However, the instability issue often causes degeneration in practice and thus attracts a lot of studies. While the traditional graph sparse coding preserves the neighborhood structure of the data, this study integrates the low-rank representation(LRR) to fix the inconsistency of sparse coding by holding the subspace structures of the high-dimensional observations. The proposed method is dubbed low-rank graph regularized sparse coding (LogSC), which learns sparse codes and low-rank representations jointly rather than the traditional two-step approach. Since the two data representations share a dictionary matrix, the resulted sparse representation on this dictionary could be benefited from LRR. We solved the optimization problem of LogSC by using the linearized alternating direction method with adaptive penalty. Experimental results show the proposed method is discriminative in feature learning and robust to various noises. This work provides a one-step approach to integrating graph embedding in representation learning. INDEX TERMS Graph sparse coding, graph embedding, data structure preserving, robust data representation, image representation, one-step integrated graph sparse coding.

Educational Data Mining Techniques for Student Performance Prediction: Method Review and Comparison Analysis

“…Degree program always requires students to take a set of courses in order, due to the knowledge provided by the previous courses being essential for subsequent courses (Tabandeh and Sami, 2010 ; Wang et al, 2015 ; Alario-Hoyos et al, 2016 ; Ren et al, 2016 ; Morsy and Karypis, 2017 ). With this idea, Polyzou et al developed course-specific regression (CSR) (Polyzou and Karypis, 2016 ), and predict student grades in a course using a sparse linear combination (Zhang and Liu, 2020 ). Following this way, there are many improved works (Polyzou and Karypis, 2016 ; Hu et al, 2017 ; Morsy and Karypis, 2017 ).…”

“…Many studies employed the traditional machine-learning methods, such as decision trees (DTs) (Al-Radaideh et al, 2006 ; Koprinska et al, 2015 ), neighborhood method (Meier et al, 2015 ), LR (Anozie and Junker, 2006 ), neural networks (Andrews et al, 1995 ; Sorour et al, 2014 ), and kernel-based method (Boser et al, 1992 ). The new feature-learning techniques have been investigated in SPP, such as Lasso regression (Sorour et al, 2014 ; Zhang et al, 2018b ; Zhang and Liu, 2020 ), matrix factorization (MF) (Slim et al, 2014 ), tensor factorization (TF) (Thai-Nghe et al, 2011a ), and deep neural networks (Kim et al, 2018 ). In these methods, MF and deep learning have attracted increasing attention for SPP.…”

A Self-Representation Weighted based Density Peaks Clustering Method

“…So far, many methods [16][17][18][19][24][25] are proposed to measure the similarities of all the data points. In these methods, sparse representation is a typical resolution and has been confirmed its outperformed performance on the high-dimensional and/or highly sparse datasets.…”

Identifying Non-Math Students from Brain MRIs with an Ensemble Classifier Based on Subspace-Enhanced Contrastive Learning

“…Yupei et al, developed the model of item response theory to predict student responses to the following questions by training the latent factor model on response records [ 4 , 12 ]. Under popular data science and technology, [ 14 , 15 ], AI-aided education tools and education discovery are becoming hot study fields, e.g., educational data mining (EDM) and learning analytics (LA) [ 4 , 5 , 12 , 16 ]. Natalia et al, presents a study of cognitive test anxiety and self-perception through questionnaires from over 2000 primary school students and 200 teachers, showing girls are more likely to experiment with a negative attitude toward mathematics than boys in Spain [ 17 ].…”