Flow3D-OPD: Multi-Teacher On-Policy Distillation for 3D Geometry Generation with Flow-Matching Diffusion Transformer
该研究通过引入多教师在线策略蒸馏方法Flow3D-OPD,解决了3D几何生成中定义奖励困难和梯度干扰问题,提升了3D模型的质量。
该研究通过引入多教师在线策略蒸馏方法Flow3D-OPD,解决了3D几何生成中定义奖励困难和梯度干扰问题,提升了3D模型的质量。
研究通过全球引用网络分析,使用网络重排零模型方法,发现中国科研的本土引用偏好低于普遍认知,并逐渐减少,且其颠覆性影响正在接近美国。
This study addresses the lack of CVE trigger assessment and configuration redundancy in Linux kernels by proposing FCC, a framework that automatically infers 1-minimal trigger configurations satisfying both build-time and runtime constraints. Integrating Kconfig constraint solving, implicit dependency completion, and topology-guided minimization, FCC is the first to generate auditable trigger boundaries validated via olddefconfig. Experimental results demonstrate that the configuration success rate increases from 62.5% to 96.6%, while the average candidate size is reduced by 78.7%. These improvements significantly lower evaluation overhead and effectively enable vendors to precisely determine CVE triggerability within specific deployments.
This work addresses the challenges of semantic-control misalignment, action inconsistency, and unreliable termination in language-guided drone navigation within target-visible environments. To this end, the authors propose DBFly, a novel framework that introduces an explicit vision-guided spatial reasoning chain prior to waypoint generation. This chain comprises target-direction anchoring, spatial diagnosis, and maneuver decision-making, complemented by implicit flight corridor modeling and a terminal convergence-aware stopping strategy to reliably bridge high-level linguistic intent with continuous low-level control. Experimental results demonstrate that DBFly achieves a 25.07 percentage point improvement in average success rate over the strongest baseline across both seen and unseen objects and scenes, significantly enhancing navigation stability and reliability.
Existing 3D point cloud generation methods struggle to simultaneously capture global topology and fine local details, often relying on computationally expensive ODE solvers or multi-step denoising procedures. This work proposes a Hierarchical Flow Matching (HFM) framework that extends flow matching into a two-level structure: it first models the global shape manifold via implicit flow matching in a compact latent space, then performs conditional point flow matching—conditioned on the learned latent code—to reconstruct detailed geometry. Both stages are trained with simple MSE regression and leverage optimal transport paths with Euler integration for efficient sampling. The method achieves state-of-the-art or comparable generation quality on ShapeNet and ModelNet, producing high-fidelity point clouds with only 15 sampling steps per level, while also yielding a structured latent space amenable to downstream tasks.
该研究通过引入多教师在线策略蒸馏方法Flow3D-OPD,解决了3D几何生成中定义奖励困难和梯度干扰问题,提升了3D模型的质量。
研究通过全球引用网络分析,使用网络重排零模型方法,发现中国科研的本土引用偏好低于普遍认知,并逐渐减少,且其颠覆性影响正在接近美国。
This study addresses the lack of CVE trigger assessment and configuration redundancy in Linux kernels by proposing FCC, a framework that automatically infers 1-minimal trigger configurations satisfying both build-time and runtime constraints. Integrating Kconfig constraint solving, implicit dependency completion, and topology-guided minimization, FCC is the first to generate auditable trigger boundaries validated via olddefconfig. Experimental results demonstrate that the configuration success rate increases from 62.5% to 96.6%, while the average candidate size is reduced by 78.7%. These improvements significantly lower evaluation overhead and effectively enable vendors to precisely determine CVE triggerability within specific deployments.
This work addresses the challenges of semantic-control misalignment, action inconsistency, and unreliable termination in language-guided drone navigation within target-visible environments. To this end, the authors propose DBFly, a novel framework that introduces an explicit vision-guided spatial reasoning chain prior to waypoint generation. This chain comprises target-direction anchoring, spatial diagnosis, and maneuver decision-making, complemented by implicit flight corridor modeling and a terminal convergence-aware stopping strategy to reliably bridge high-level linguistic intent with continuous low-level control. Experimental results demonstrate that DBFly achieves a 25.07 percentage point improvement in average success rate over the strongest baseline across both seen and unseen objects and scenes, significantly enhancing navigation stability and reliability.
Existing 3D point cloud generation methods struggle to simultaneously capture global topology and fine local details, often relying on computationally expensive ODE solvers or multi-step denoising procedures. This work proposes a Hierarchical Flow Matching (HFM) framework that extends flow matching into a two-level structure: it first models the global shape manifold via implicit flow matching in a compact latent space, then performs conditional point flow matching—conditioned on the learned latent code—to reconstruct detailed geometry. Both stages are trained with simple MSE regression and leverage optimal transport paths with Euler integration for efficient sampling. The method achieves state-of-the-art or comparable generation quality on ShapeNet and ModelNet, producing high-fidelity point clouds with only 15 sampling steps per level, while also yielding a structured latent space amenable to downstream tasks.