The feature-extraction quality of pre-trained models sets the ceiling of downstream transfer learning, yet few-shot model selection remains difficult: supervised evaluation on scarce labels is unstable and yields fluctuating rankings. We propose Entropy-Guided Model Evaluation (EGiME), a semi-supervised criterion that...
De-Quan Jin, Tao Li, Li-Feng Xing et al.· 0 citations
This article elaborate on how HGNNs address the limitations of traditional graph-based models by leveraging hyperedge-based aggregation mechanisms to encode higher-order semantics, and highlights their versatile applications across interdisciplinary fields, including drug discovery, protein interaction prediction, visu...
Yue Gao, Yi-Fan Feng, Xiangmin Han et al.· IEEE Transactions on Pattern...· 0 citations
A unified Latent Initialization and Gradient Optimization (LINGO) framework that combines latent mask-space initialization with dynamic gradient optimization to improve point cloud structural completeness while accelerating training and enhancing reconstruction quality across different sparse-view imaging scenarios is...
Li-Feng Xing, De-Quan Jin, Kunpeng Bu et al.· arXiv.org· 1 citation
Hyper-Fold is introduced, a rank-K separable convolutional backbone approaching this ceiling at message-passing cost, suggesting that a sufficiently expressive 3D backbone recovers information that fusion architectures previously borrowed from evolution-scale pretraining.
Yifan Feng, Guang Cheng, Shihui Ying et al.· 0 citations
The Hypergraph Identity-Aware Subtree (IA Subtree) Kernel is introduced, which distinguishes uniform-regular hypergraphs by considering both neighborhood connectivity and connection density and develops two Hypergraph Neural Networks: Hypergraph Isomorphism Networks (HGIN) and Identity-Aware Hypergraph Isomorphism Netw...
Yi-Fan Feng, Rizhuo Huang, Yi-Fan Zhang et al.· IEEE Transactions on Pattern...· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.