Selected work

Representative Papers

Neuro-Bayesian-Symbolic Residual Attention Shallow Network: Explainable Deep Learning for Cybersecurity Risk Assessment

Jun 29, 2026

This study addresses the need for trustworthy cybersecurity assessment in high-risk scenarios within open-source ecosystems by proposing a shallow, interpretable hybrid architecture that integrates neural networks, Bayesian inference, and symbolic logic. The approach embeds domain knowledge and causal reasoning through differentiable components, augmented with a residual attention mechanism and expert-defined risk amplifiers—such as blast radius and propagation velocity—to emulate deep reasoning while preserving structural transparency. Innovatively, five epistemic axioms—precision, causality, falsifiability, transparency, and completeness—are hard-constrained into the gating mechanism to guarantee intrinsic interpretability. Evaluated on 20 open-source projects covering all OWASP Top 10:2025 categories and multilingual vulnerabilities, the model achieves confidence scores of 0.79–0.97 and outperforms conventional black-box models in accuracy.

0 citationsRead paper

Three Barriers to Kantian Cooperation under Inequality

Jun 20, 2026

This study investigates the coordination and distributional barriers to voluntary adoption of Kantian cooperation under inequality. Building on a two-agent voluntary contribution model with logarithmic utility, it compares Nash equilibria and multiplicative Kantian equilibria in both contribution and private consumption spaces. The analysis systematically identifies and reconstructs three distinct obstacles that arise when transitioning from Nash to Kantian equilibria: non-Pareto-improving outcomes, allocation conflicts induced by strategic space parameterization, and coordination difficulties along a continuum of Pareto-efficient allocations. The findings reveal that under high inequality, the poorer agent may prefer the Nash outcome; alternative parameterizations generate divergent distributional implications; and the continuity of the Pareto frontier intensifies coordination challenges. Collectively, these results delineate theoretical boundaries for the feasibility and sustainability of Kantian cooperation.

0 citationsRead paper

Moral Geometry: Endogenous Scaling in Nash-Kantian Games

Jun 20, 2026

This study addresses the strategic vulnerability of Kantian equilibria under multiplicative transformations, which lack invariance to monotonic rescalings of strategies and thereby place Kantian agents at a disadvantage. To resolve this, the paper introduces an endogenous, smoothly increasing scaling function as a commitment device that redefines deviation assessments without altering material payoffs or feasible sets. Through game-theoretic analysis, elasticity-weighted modeling, and construction of local elasticity ratios—illustrated via Cournot and differentiated Bertrand models—the mechanism yields distinct outcomes: under strategic complementarity, it strictly outperforms the Nash equilibrium and attains Stackelberg leader payoffs; under strategic substitutability, it effectively avoids the payoff losses associated with naive Kantian behavior, approaching but never exceeding the opponent’s Nash payoff, thus exhibiting a dual offensive-defensive character.

0 citationsRead paper

Bernoulli CUSUM and Bayes-Optimal Detection Ceilings for Trust Fraud in Sparse Rating Networks

Jun 03, 2026

Existing continuous observation models fail in sparse binary rating networks—such as Bitcoin-OTC—due to data discretization, hindering effective trust fraud detection. This work proposes the first sequential detection framework tailored for such networks, combining an accurate parametric model derived from empirical distributions with a dual-mechanism architecture that integrates Bernoulli CUSUM and asymmetric rating modulation to jointly capture behavioral shifts and rating dynamics. Theoretically, we derive a Bayesian-optimal F1 upper bound based on empirical parameters and demonstrate that detector–model alignment outweighs raw information quantity in detection efficacy. Experiments on Bitcoin-OTC and Bitcoin-Alpha achieve AUC scores of 0.749 and 0.796, respectively, significantly outperforming GaaSTrust (p < 0.003), with founder-labeled instances attaining an AUC of 0.999.

0 citationsRead paper

The Full Pareto Frontier as Kantian Equilibria

May 19, 2026

This study addresses the limitation of Multiplicative Kantian equilibria in failing to cover the Pareto frontier due to constraints imposed by the parameterization of the strategy space. To overcome this, the authors propose a coordinate translation—specifically, imposing lower bounds on actions—to reconstruct the strategy space. Through geometric analysis leveraging the common tangent property of players’ indifference curves, they constructively demonstrate that for any interior Pareto-efficient outcome, there exists a translation under which it becomes a Multiplicative Kantian equilibrium. This result establishes, for the first time, that such equilibria can span the entire set of interior points on the Pareto frontier, thereby decoupling efficiency from fairness: any normative criterion can be implemented without compromising Pareto optimality. This significantly broadens the theoretical applicability and explanatory power of Kantian equilibria.

0 citationsRead paper

Recent publications

Latest Papers

Neuro-Bayesian-Symbolic Residual Attention Shallow Network: Explainable Deep Learning for Cybersecurity Risk Assessment

Jun 29, 2026

This study addresses the need for trustworthy cybersecurity assessment in high-risk scenarios within open-source ecosystems by proposing a shallow, interpretable hybrid architecture that integrates neural networks, Bayesian inference, and symbolic logic. The approach embeds domain knowledge and causal reasoning through differentiable components, augmented with a residual attention mechanism and expert-defined risk amplifiers—such as blast radius and propagation velocity—to emulate deep reasoning while preserving structural transparency. Innovatively, five epistemic axioms—precision, causality, falsifiability, transparency, and completeness—are hard-constrained into the gating mechanism to guarantee intrinsic interpretability. Evaluated on 20 open-source projects covering all OWASP Top 10:2025 categories and multilingual vulnerabilities, the model achieves confidence scores of 0.79–0.97 and outperforms conventional black-box models in accuracy.

0 citationsRead paper

Three Barriers to Kantian Cooperation under Inequality

Jun 20, 2026

This study investigates the coordination and distributional barriers to voluntary adoption of Kantian cooperation under inequality. Building on a two-agent voluntary contribution model with logarithmic utility, it compares Nash equilibria and multiplicative Kantian equilibria in both contribution and private consumption spaces. The analysis systematically identifies and reconstructs three distinct obstacles that arise when transitioning from Nash to Kantian equilibria: non-Pareto-improving outcomes, allocation conflicts induced by strategic space parameterization, and coordination difficulties along a continuum of Pareto-efficient allocations. The findings reveal that under high inequality, the poorer agent may prefer the Nash outcome; alternative parameterizations generate divergent distributional implications; and the continuity of the Pareto frontier intensifies coordination challenges. Collectively, these results delineate theoretical boundaries for the feasibility and sustainability of Kantian cooperation.

0 citationsRead paper

Moral Geometry: Endogenous Scaling in Nash-Kantian Games

Jun 20, 2026

This study addresses the strategic vulnerability of Kantian equilibria under multiplicative transformations, which lack invariance to monotonic rescalings of strategies and thereby place Kantian agents at a disadvantage. To resolve this, the paper introduces an endogenous, smoothly increasing scaling function as a commitment device that redefines deviation assessments without altering material payoffs or feasible sets. Through game-theoretic analysis, elasticity-weighted modeling, and construction of local elasticity ratios—illustrated via Cournot and differentiated Bertrand models—the mechanism yields distinct outcomes: under strategic complementarity, it strictly outperforms the Nash equilibrium and attains Stackelberg leader payoffs; under strategic substitutability, it effectively avoids the payoff losses associated with naive Kantian behavior, approaching but never exceeding the opponent’s Nash payoff, thus exhibiting a dual offensive-defensive character.

0 citationsRead paper

Bernoulli CUSUM and Bayes-Optimal Detection Ceilings for Trust Fraud in Sparse Rating Networks

Jun 03, 2026

Existing continuous observation models fail in sparse binary rating networks—such as Bitcoin-OTC—due to data discretization, hindering effective trust fraud detection. This work proposes the first sequential detection framework tailored for such networks, combining an accurate parametric model derived from empirical distributions with a dual-mechanism architecture that integrates Bernoulli CUSUM and asymmetric rating modulation to jointly capture behavioral shifts and rating dynamics. Theoretically, we derive a Bayesian-optimal F1 upper bound based on empirical parameters and demonstrate that detector–model alignment outweighs raw information quantity in detection efficacy. Experiments on Bitcoin-OTC and Bitcoin-Alpha achieve AUC scores of 0.749 and 0.796, respectively, significantly outperforming GaaSTrust (p < 0.003), with founder-labeled instances attaining an AUC of 0.999.

0 citationsRead paper

The Full Pareto Frontier as Kantian Equilibria

May 19, 2026

This study addresses the limitation of Multiplicative Kantian equilibria in failing to cover the Pareto frontier due to constraints imposed by the parameterization of the strategy space. To overcome this, the authors propose a coordinate translation—specifically, imposing lower bounds on actions—to reconstruct the strategy space. Through geometric analysis leveraging the common tangent property of players’ indifference curves, they constructively demonstrate that for any interior Pareto-efficient outcome, there exists a translation under which it becomes a Multiplicative Kantian equilibrium. This result establishes, for the first time, that such equilibria can span the entire set of interior points on the Pareto frontier, thereby decoupling efficiency from fairness: any normative criterion can be implemented without compromising Pareto optimality. This significantly broadens the theoretical applicability and explanatory power of Kantian equilibria.

0 citationsRead paper