Selected work

Representative Papers

On Asynchrony and Reversibility in CCS

Sep 25, 2026Electronic Proceedings in Theoretical Computer Science

This study addresses the theoretical gap in reversible process calculi, which are largely confined to synchronous settings, by investigating the integration of asynchronous communication with reversible computation. We propose CCSa, an asynchronous variant of CCS, along with its reversible extension rCCSa. Building upon the Phillips-Ulidowski framework and Lanese’s axiomatic system, our approach employs unique key annotations to explicitly model the generation and consumption of message events, thereby enabling computation rollback that preserves causal dependencies. This work makes a pioneering contribution by introducing reversible semantics into asynchronous CCS, establishing a causally consistent reversible model. Furthermore, we rigorously prove that rCCSa satisfies causal consistency, ensuring that computations can be precisely rolled back to causally equivalent states. These results fill a critical theoretical void in the field of reversible concurrent computation.

0 citationsRead paper

A Simple Gripper Interface for Simulator-Agnostic Cloth Manipulation

Sep 24, 2026

This study addresses the challenges of complex grasping models, reliance on frictional contact, and difficult cross-simulator transfer in cloth manipulation by proposing a simplified position-constraint-based grasping model. By eliminating complex friction modeling, the method defines grasp regions using axis-aligned bounding boxes and pyramidal volumes, achieving simulator-agnostic control through the selection and transmission of cloth vertices. Integrating discrete position constraints, stretch-free simulation, and a projection-based solver, it supports progressive squeezing to smooth motion trajectories. Experimental results demonstrate that the proposed model exhibits strong robustness across diverse simulation environments and successfully guides a physical robotic manipulator in completing cloth folding tasks.

0 citationsRead paper

Deep Image Segmentation via Discriminant Feature Learning

May 14, 2026

This work addresses the limitations of existing image segmentation methods, which often produce blurry boundaries and low-confidence predictions due to loss functions that neglect the discriminative structure of features. To overcome this, we propose a differentiable, architecture-agnostic Deep Discriminant Analysis (DDA) loss that integrates classical discriminant analysis principles into end-to-end training. The DDA loss explicitly maximizes inter-class variance while minimizing intra-class variance, thereby encouraging the learning of compact and well-separated feature representations. Notably, it incurs no additional inference overhead and can be seamlessly incorporated into any segmentation network. Extensive experiments on the DIS5K benchmark demonstrate that DDA consistently enhances segmentation accuracy, boundary sharpness, and model confidence, confirming the effectiveness of discriminative feature learning for image segmentation.

0 citationsRead paper

Recent publications

Latest Papers

On Asynchrony and Reversibility in CCS

Sep 25, 2026Electronic Proceedings in Theoretical Computer Science

This study addresses the theoretical gap in reversible process calculi, which are largely confined to synchronous settings, by investigating the integration of asynchronous communication with reversible computation. We propose CCSa, an asynchronous variant of CCS, along with its reversible extension rCCSa. Building upon the Phillips-Ulidowski framework and Lanese’s axiomatic system, our approach employs unique key annotations to explicitly model the generation and consumption of message events, thereby enabling computation rollback that preserves causal dependencies. This work makes a pioneering contribution by introducing reversible semantics into asynchronous CCS, establishing a causally consistent reversible model. Furthermore, we rigorously prove that rCCSa satisfies causal consistency, ensuring that computations can be precisely rolled back to causally equivalent states. These results fill a critical theoretical void in the field of reversible concurrent computation.

0 citationsRead paper

A Simple Gripper Interface for Simulator-Agnostic Cloth Manipulation

Sep 24, 2026

This study addresses the challenges of complex grasping models, reliance on frictional contact, and difficult cross-simulator transfer in cloth manipulation by proposing a simplified position-constraint-based grasping model. By eliminating complex friction modeling, the method defines grasp regions using axis-aligned bounding boxes and pyramidal volumes, achieving simulator-agnostic control through the selection and transmission of cloth vertices. Integrating discrete position constraints, stretch-free simulation, and a projection-based solver, it supports progressive squeezing to smooth motion trajectories. Experimental results demonstrate that the proposed model exhibits strong robustness across diverse simulation environments and successfully guides a physical robotic manipulator in completing cloth folding tasks.

0 citationsRead paper

Deep Image Segmentation via Discriminant Feature Learning

May 14, 2026

This work addresses the limitations of existing image segmentation methods, which often produce blurry boundaries and low-confidence predictions due to loss functions that neglect the discriminative structure of features. To overcome this, we propose a differentiable, architecture-agnostic Deep Discriminant Analysis (DDA) loss that integrates classical discriminant analysis principles into end-to-end training. The DDA loss explicitly maximizes inter-class variance while minimizing intra-class variance, thereby encouraging the learning of compact and well-separated feature representations. Notably, it incurs no additional inference overhead and can be seamlessly incorporated into any segmentation network. Extensive experiments on the DIS5K benchmark demonstrate that DDA consistently enhances segmentation accuracy, boundary sharpness, and model confidence, confirming the effectiveness of discriminative feature learning for image segmentation.

0 citationsRead paper