Selected work

Representative Papers

No One to Blame: A Framework of Constitutive AI Unaccountability

Aug 12, 2026

This study addresses the structural accountability gaps engendered by autonomous AI systems, which elude conventional accountability mechanisms. It introduces the concept of “constitutive AI unaccountability,” framing unaccountability not as a failure to be remedied but as an inherent property of sociotechnical systems, thereby extending existing theories of accountability barriers. Through a three-phase qualitative investigation—comprising concept-driven literature analysis, secondary analysis of interviews with 27 experts, and application of the emerging framework to the open-source system OpenClaw—the research identifies nine categories encompassing twenty constitutive unaccountability themes and maps eight interdependencies among them. The authors develop an operational diagnostic instrument comprising twenty targeted questions, which successfully detected seventeen unaccountability conditions in OpenClaw, offering a novel paradigm for evaluating accountability in AI systems.

0 citationsRead paper

Weak Bisimulation Finiteness of Pushdown Systems With Deterministic $\varepsilon$-Transitions Is 2-ExpTime-Complete

Aug 11, 2026

This study addresses whether pushdown systems with deterministic ε-transitions are weakly bisimilar to some finite-state system. By integrating formal verification, automata theory, and complexity analysis, we precisely characterize the computational complexity of this decision problem as 2-ExpTime-complete: we present a 2-ExpTime decision algorithm and establish matching 2-ExpTime-hardness. Furthermore, we prove that the size of the minimal equivalent finite-state system is at most doubly exponential in the size of the input. This result substantially improves upon the previous gap between an Ackermannian upper bound and an ExpTime lower bound, providing a tight complexity classification for the problem.

0 citationsRead paper

When Attention Goes Blind: Numerical Failure in ALiBi Positional Encodings

Aug 04, 2026

This work identifies a critical numerical instability in the Attention with Linear Biases (ALiBi) positional encoding, wherein linear bias scaling causes underflow in floating-point precision, leading to excessive zeroing of attention weights and effectively “blinding” attention heads. The study is the first to characterize this failure mechanism, systematically evaluates its impact on long-context modeling, and introduces four training-time mitigation strategies, including a novel log-distance scaling approach. Experiments on a 148M-parameter decoder demonstrate that standard ALiBi remains a strong baseline on needle-in-a-haystack tasks, while the proposed log-scaled distance consistently enhances performance—particularly in passkey retrieval—significantly improving token retrieval capability and offering clear practical guidance for robust ALiBi training.

0 citationsRead paper

Node Labeling in Line Diagrams of Ordered Sets

Jul 25, 2026

This study addresses the challenge of label overlap with nodes, edges, and other labels in ordered parallel coordinates plots by proposing a two-stage flexible labeling algorithm that balances layout readability with minimal connection distance. The method first pre-selects fixed-position labels based on readability criteria; for overflow labels, it leverages both internal and external plot space to connect them to their corresponding nodes via non-crossing straight lines. An initial placement is computed using a sparse grid-based cost function, followed by refinement through a force-directed model. Notably, this approach introduces the first strategy specifically tailored to the vertical-edge characteristics of ordered parallel coordinates, flexibly accommodating the dual-region constraints inherent in formal concept analysis and requiring only minor adjustments to suit concept lattice labeling. Experimental results demonstrate that the algorithm efficiently produces non-crossing, highly readable label layouts and has been successfully applied to concept lattice visualization.

0 citationsRead paper

Performance Analysis in Parallel Programming Education: A Comparative Usability Study

Jun 30, 2026

This work addresses the high complexity of existing high-performance computing (HPC) performance analysis tools, which hinders students’ intuitive understanding of parallel program performance issues. To bridge this gap, the paper introduces EduMPI—the first educational tool that integrates HPC cluster operations and MPI performance analysis within a streamlined graphical interface. EduMPI enables near real-time, physically node-layout-aware communication visualization, facilitating interactive identification of load imbalance and other performance bottlenecks. User studies demonstrate that, compared to professional-grade tools, EduMPI significantly lowers the learning barrier and effectively enhances students’ comprehension of parallel performance characteristics, thereby improving the practicality and accessibility of parallel programming education.

0 citationsRead paper

Recent publications

Latest Papers

No One to Blame: A Framework of Constitutive AI Unaccountability

Aug 12, 2026

This study addresses the structural accountability gaps engendered by autonomous AI systems, which elude conventional accountability mechanisms. It introduces the concept of “constitutive AI unaccountability,” framing unaccountability not as a failure to be remedied but as an inherent property of sociotechnical systems, thereby extending existing theories of accountability barriers. Through a three-phase qualitative investigation—comprising concept-driven literature analysis, secondary analysis of interviews with 27 experts, and application of the emerging framework to the open-source system OpenClaw—the research identifies nine categories encompassing twenty constitutive unaccountability themes and maps eight interdependencies among them. The authors develop an operational diagnostic instrument comprising twenty targeted questions, which successfully detected seventeen unaccountability conditions in OpenClaw, offering a novel paradigm for evaluating accountability in AI systems.

0 citationsRead paper

Weak Bisimulation Finiteness of Pushdown Systems With Deterministic $\varepsilon$-Transitions Is 2-ExpTime-Complete

Aug 11, 2026

This study addresses whether pushdown systems with deterministic ε-transitions are weakly bisimilar to some finite-state system. By integrating formal verification, automata theory, and complexity analysis, we precisely characterize the computational complexity of this decision problem as 2-ExpTime-complete: we present a 2-ExpTime decision algorithm and establish matching 2-ExpTime-hardness. Furthermore, we prove that the size of the minimal equivalent finite-state system is at most doubly exponential in the size of the input. This result substantially improves upon the previous gap between an Ackermannian upper bound and an ExpTime lower bound, providing a tight complexity classification for the problem.

0 citationsRead paper

When Attention Goes Blind: Numerical Failure in ALiBi Positional Encodings

Aug 04, 2026

This work identifies a critical numerical instability in the Attention with Linear Biases (ALiBi) positional encoding, wherein linear bias scaling causes underflow in floating-point precision, leading to excessive zeroing of attention weights and effectively “blinding” attention heads. The study is the first to characterize this failure mechanism, systematically evaluates its impact on long-context modeling, and introduces four training-time mitigation strategies, including a novel log-distance scaling approach. Experiments on a 148M-parameter decoder demonstrate that standard ALiBi remains a strong baseline on needle-in-a-haystack tasks, while the proposed log-scaled distance consistently enhances performance—particularly in passkey retrieval—significantly improving token retrieval capability and offering clear practical guidance for robust ALiBi training.

0 citationsRead paper

Node Labeling in Line Diagrams of Ordered Sets

Jul 25, 2026

This study addresses the challenge of label overlap with nodes, edges, and other labels in ordered parallel coordinates plots by proposing a two-stage flexible labeling algorithm that balances layout readability with minimal connection distance. The method first pre-selects fixed-position labels based on readability criteria; for overflow labels, it leverages both internal and external plot space to connect them to their corresponding nodes via non-crossing straight lines. An initial placement is computed using a sparse grid-based cost function, followed by refinement through a force-directed model. Notably, this approach introduces the first strategy specifically tailored to the vertical-edge characteristics of ordered parallel coordinates, flexibly accommodating the dual-region constraints inherent in formal concept analysis and requiring only minor adjustments to suit concept lattice labeling. Experimental results demonstrate that the algorithm efficiently produces non-crossing, highly readable label layouts and has been successfully applied to concept lattice visualization.

0 citationsRead paper

Performance Analysis in Parallel Programming Education: A Comparative Usability Study

Jun 30, 2026

This work addresses the high complexity of existing high-performance computing (HPC) performance analysis tools, which hinders students’ intuitive understanding of parallel program performance issues. To bridge this gap, the paper introduces EduMPI—the first educational tool that integrates HPC cluster operations and MPI performance analysis within a streamlined graphical interface. EduMPI enables near real-time, physically node-layout-aware communication visualization, facilitating interactive identification of load imbalance and other performance bottlenecks. User studies demonstrate that, compared to professional-grade tools, EduMPI significantly lowers the learning barrier and effectively enhances students’ comprehension of parallel performance characteristics, thereby improving the practicality and accessibility of parallel programming education.

0 citationsRead paper