Hypergraph based dual constraint propagation robust semi-supervised nonnegative matrix factorization for image clustering

Key Points

• Improved image clustering effectiveness is achieved using a novel nonnegative matrix factorization technique that incorporates hypergraph design.
• The dual constraint propagation method significantly enhances the algorithm's robustness, leading to better clustering outcomes.
• Observational analysis reveals that this semi-supervised approach allows for more accurate clustering with less labeled data required.
• The findings highlight the potential for using advanced hypergraph methods to push the boundaries of current clustering techniques.