This work designed and implemented an embodied intelligent skill orchestration framework (A-Harness) based on verification gating. Through a closed-loop verification mechanism and dual-layer memory enhancement, it significantly reduces inference costs while achieving high-precision interaction point localization in an open world.

This work proposes a cost-aware framework (AgentTailor) that evaluates the semantic contribution of communication edges through an edge-judgment mechanism and uses an edge prediction network (EPN) to estimate the utility of edges through virtual execution.

This work introduces a structure-aware prototype learning mechanism and develops a prototype-guided fine-grained fusion strategy.

This work proposes a parameterless dual-granularity weighted multi-view fuzzy c-means clustering framework and a parameterless two-layer local weighted clustering algorithm.

This work proposes a nonparametric two-layer locally weighted k-means (NDL-WKM) method that can achieve adaptive and fine-grained clustering without any additional parameters.

This work proposes SESM divergence based on evidence theory and designs a multi-sensor data fusion method based on SESM divergence and Deng entropy.

This work proposes a self-weighted multi-view intelligent fuzzy clustering (AW-MVNFC) algorithm, which adaptively learns the contribution of each data view and the importance of each feature in each view through a hierarchical weighting strategy.

This work proposes a multi-view K-means clustering framework based on Exponential Saturation distance and Hierarchical weight learning.