Selected work

Representative Papers

Bayesian updates from coalgebraic determinisation

Jun 24, 2026

This work addresses the limitation of traditional POMDP determinization methods, which discard intermediate observations and thus fail to support full-history Bayesian updating. By integrating unifilarisation into a coalgebraic determinization framework, the authors introduce a support structure over a monoid that represents system states as prior distributions and defines transitions via Bayesian filtering. They establish, for the first time, that unifilarisation is a special case of coalgebraic determinization, thereby naturally embedding Bayesian updating within a categorical semantics. This approach extends to stochastic Mealy machines equipped with support structures, yielding a semantics finer than conventional Moore models. The resulting framework generates, for each input word, a family of output distributions satisfying causal constraints, making it well-suited for modeling reinforcement learning and sequential decision-making problems.

0 citationsRead paper

Compositional Behavioral Semantics for State Abstraction in Reinforcement Learning

Jun 23, 2026

Existing state abstraction methods lack a general principle for rigorously preserving behavioral structure. This work proposes a unified framework that defines behavioral semantics in reinforcement learning in a compositional manner, grounded in local one-step descriptions of system dynamics, and establishes a theory for safe transfer of behavioral structure between abstract and concrete systems. For the first time, the framework enables a compositional formalization of behavioral semantics, supporting the derivation of quantitative metrics with correctness guarantees from logical semantics. It thus lays a principled foundation for behavioral reasoning under state abstraction and provides reusable definitions and provably faithful transfer mechanisms applicable to a broad class of behavioral structures.

0 citationsRead paper

An Augmented Reality Brain-Robot Interface for Generalist Robot Arm Manipulation

Jun 15, 2026

This study addresses the limited generalizability of existing augmented reality (AR) brain–computer interface (BCI) systems, which are often confined to specific tasks and struggle to support versatile robotic manipulation in real-world settings. The authors propose a novel AR-based shared autonomy framework that integrates eye tracking with motor imagery electroencephalography (EEG) signals: users select targets via gaze, while the system leverages contextual visual cues—such as “place” and “use”—to guide multi-step everyday tasks. This work represents the first integration of AR, eye tracking, and EEG-based BCI for general-purpose robotic arm control, overcoming task-specific constraints to enable intuitive, sequential interaction. In experiments with 18 participants, the system successfully facilitated complex activities including drinking, opening drawers, and operating an oven, achieving strong usability (System Usability Scale > 70) and demonstrating the feasibility and promise of this paradigm for BCI-driven robotic assistance.

0 citationsRead paper

Intrinsic Computational Functionalism and Simulated Consciousness

Jun 13, 2026

This study addresses the objection that simulated consciousness is unreal by proposing, within the framework of internalist computational functionalism, an Intrinsic Causal Computational Realization (ICCR) relation to characterize the intrinsic causal-computational organization on which consciousness depends. Departing from traditional approaches that focus solely on input–output behavior, this work integrates normative functionalism, causal intervention theory, and computational structural modeling to introduce internal mechanisms, intervention operations, and joint readouts, thereby establishing a criterion for consciousness equivalence that accounts for both physical implementation and intrinsic state structure. The findings suggest that if consciousness is an invariant of intrinsic causal-computational organization, then biological, artificial, or simulated systems satisfying ICCR all realize identical conscious properties.

0 citationsRead paper

Deep Sleep Classification via EEG Signal Criticality: A Passive BCI Approach for Sleep-Improvement Neurofeedback

Jun 11, 2026

This study proposes a high-accuracy automatic staging method for deep sleep (N3 stage) based on criticality features derived from electroencephalography (EEG), aimed at enabling intention-free closed-loop neurofeedback interventions. For the first time, criticality indices extracted via detrended fluctuation analysis (DFA) are employed for deep sleep identification. The approach integrates UMAP manifold learning with multiple classifiers and is validated on 347,232 EEG segments from 290 older women. Results demonstrate that Naïve Bayes significantly outperforms both deep neural networks and random forests in this nonlinear manifold task, achieving a mean balanced accuracy of 87.17%. This work establishes a novel paradigm for state-dependent neurofeedback by leveraging EEG criticality as a robust biomarker for deep sleep detection.

0 citationsRead paper

Recent publications

Latest Papers

Bayesian updates from coalgebraic determinisation

Jun 24, 2026

This work addresses the limitation of traditional POMDP determinization methods, which discard intermediate observations and thus fail to support full-history Bayesian updating. By integrating unifilarisation into a coalgebraic determinization framework, the authors introduce a support structure over a monoid that represents system states as prior distributions and defines transitions via Bayesian filtering. They establish, for the first time, that unifilarisation is a special case of coalgebraic determinization, thereby naturally embedding Bayesian updating within a categorical semantics. This approach extends to stochastic Mealy machines equipped with support structures, yielding a semantics finer than conventional Moore models. The resulting framework generates, for each input word, a family of output distributions satisfying causal constraints, making it well-suited for modeling reinforcement learning and sequential decision-making problems.

0 citationsRead paper

Compositional Behavioral Semantics for State Abstraction in Reinforcement Learning

Jun 23, 2026

Existing state abstraction methods lack a general principle for rigorously preserving behavioral structure. This work proposes a unified framework that defines behavioral semantics in reinforcement learning in a compositional manner, grounded in local one-step descriptions of system dynamics, and establishes a theory for safe transfer of behavioral structure between abstract and concrete systems. For the first time, the framework enables a compositional formalization of behavioral semantics, supporting the derivation of quantitative metrics with correctness guarantees from logical semantics. It thus lays a principled foundation for behavioral reasoning under state abstraction and provides reusable definitions and provably faithful transfer mechanisms applicable to a broad class of behavioral structures.

0 citationsRead paper

An Augmented Reality Brain-Robot Interface for Generalist Robot Arm Manipulation

Jun 15, 2026

This study addresses the limited generalizability of existing augmented reality (AR) brain–computer interface (BCI) systems, which are often confined to specific tasks and struggle to support versatile robotic manipulation in real-world settings. The authors propose a novel AR-based shared autonomy framework that integrates eye tracking with motor imagery electroencephalography (EEG) signals: users select targets via gaze, while the system leverages contextual visual cues—such as “place” and “use”—to guide multi-step everyday tasks. This work represents the first integration of AR, eye tracking, and EEG-based BCI for general-purpose robotic arm control, overcoming task-specific constraints to enable intuitive, sequential interaction. In experiments with 18 participants, the system successfully facilitated complex activities including drinking, opening drawers, and operating an oven, achieving strong usability (System Usability Scale > 70) and demonstrating the feasibility and promise of this paradigm for BCI-driven robotic assistance.

0 citationsRead paper

Intrinsic Computational Functionalism and Simulated Consciousness

Jun 13, 2026

This study addresses the objection that simulated consciousness is unreal by proposing, within the framework of internalist computational functionalism, an Intrinsic Causal Computational Realization (ICCR) relation to characterize the intrinsic causal-computational organization on which consciousness depends. Departing from traditional approaches that focus solely on input–output behavior, this work integrates normative functionalism, causal intervention theory, and computational structural modeling to introduce internal mechanisms, intervention operations, and joint readouts, thereby establishing a criterion for consciousness equivalence that accounts for both physical implementation and intrinsic state structure. The findings suggest that if consciousness is an invariant of intrinsic causal-computational organization, then biological, artificial, or simulated systems satisfying ICCR all realize identical conscious properties.

0 citationsRead paper

Deep Sleep Classification via EEG Signal Criticality: A Passive BCI Approach for Sleep-Improvement Neurofeedback

Jun 11, 2026

This study proposes a high-accuracy automatic staging method for deep sleep (N3 stage) based on criticality features derived from electroencephalography (EEG), aimed at enabling intention-free closed-loop neurofeedback interventions. For the first time, criticality indices extracted via detrended fluctuation analysis (DFA) are employed for deep sleep identification. The approach integrates UMAP manifold learning with multiple classifiers and is validated on 347,232 EEG segments from 290 older women. Results demonstrate that Naïve Bayes significantly outperforms both deep neural networks and random forests in this nonlinear manifold task, achieving a mean balanced accuracy of 87.17%. This work establishes a novel paradigm for state-dependent neurofeedback by leveraging EEG criticality as a robust biomarker for deep sleep detection.

0 citationsRead paper