Anchor-based methods have demonstrated significant success in multi-view clustering, particularly in handling large-scale datasets. However, existing approaches suffer from two critical limitations: (1) clustering performance is highly sensitive to the quality of initial anchors, and (2) multi-view interactions are often limited to the anchor graph construction stage, resulting in semantic misalignment among independently generated anchors. To address these challenges, we propose a unified paradigm based on Cross-View Semantic Propagation (CVSP), a plug-and-play framework that effectively improves anchor quality. Depending on whether the initial anchors are fixed or learnable, this strategy is instantiated in two variants, CVSP-F for fixed anchors and CVSP-U for learnable anchors with iterative updates. Specifically, the anchor alignment module first ensures cross-view consistency among the initial anchors. Then, the anchor refinement module, leveraging a cross-view graph, optimizes the anchor representations. Finally, a revised anchor graph construction module is introduced to further improve clustering robustness. Experimental results demonstrate that both CVSP-F and CVSP-U consistently outperform twelve state-of-the-art anchor-based methods across multiple datasets. Furthermore, extensive evaluations confirm the generalization capability of the anchor refinement module when integrated with various existing methods.
Suyuan Liu, Siwei Wang, Ke Liang et al.· IEEE Transactions on Pattern...· 0 citations
Reinforcement learning (RL) search agents commonly model retrieval as free-form natural-language query generation and optimize multi-turn interactions using final-answer rewards. Current studies mainly improve training with denser or more structured credit signals, but rarely examine whether retrieval is properly formulated at the policy-environment interface. We observe pronounced retrieval aliasing during Search-R1 training: rollouts for the same question continue to generate distinct query strings, yet their accumulated evidence sets increasingly overlap. We call this phenomenon retrieval-equivalence collapse; in this regime, trajectories approach utility equivalence with respect to retrieval decisions, leaving within-group returns with little effective retrieval contrast. To address this problem, we propose Harness-G, a graph-structured retrieval framework that redesigns this interface. It reformulates free-form query generation as finite action selection: the policy selects an evidence sentence or entity, or chooses to answer, while the environment constructs the menu, tracks retrieval state, and validates and executes each choice. This interface reduces linguistic aliasing and makes same-state alternatives directly comparable. Building on this interface, we introduce Structured Non-myopic Credit (SNC), which uses a frozen answer scorer to compare the selected action with its alternatives and assigns downstream gains to the earlier actions that enabled them. Across six QA benchmarks, Harness-G achieves the highest average F1 at both evaluated model scales, outperforming the strongest baseline, Graph-R1, by 10.74 points at 1.5B and 3.98 points at 3B.
Yanning Hou, Hao-Yuan Chen, Sihang Zhou et al.· 0 citations
