Pseudo-label-based deep clustering has emerged as a compelling paradigm for unsupervised visual learning, yet two fundamental limitations impede its effectiveness: (1) pseudo-label noise, where unreliable hard assignments generated in early training accumulate and propagate errors across iterations; and (2) confidence-...
Chu-Lei Zhang, Rui-Kai Liu, Yuan-Shu Li et al.· Big Data and Cognitive Compu...· 0 citations
A graph-constrained diffusion model that replaces the topology-agnostic uniform transition matrix with an instance-dependent graph Laplacian-based heat-kernel transition matrix, and a progressive distillation framework that iteratively halves the required denoising steps via teacher–student parameter inheritance is int...
Chu-Lei Zhang, Yun-Shuo Li, Fu-Shuo Li et al.· Big Data and Cognitive Compu...· 0 citations
The proposed ReliableRAG is the first reliability-driven framework that mitigates deceptive misinformation in multi-hop QA through fine-grained evaluation of individual triples, and quantifies triple reliability by combining query-triple semantic relevance with triple credibility.
Jingjing Jiang, Xuan Wu, Wen-Hao Song et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.