Selected work

Representative Papers

Deep Learning Methods in Neuroscience: From Modeling Molecular Mechanisms to Classifying States of Consciousness

Sep 28, 2026

This study addresses the limited model interpretability and lack of standardization in classifying states of consciousness and modeling anesthetized brain dynamics by proposing a hybrid, generalizable architecture that integrates physiological priors. By combining deep neural networks, multiscale computational modeling, and clustering algorithms, this work jointly analyzes EEG, fMRI, and LFP data to achieve automated brain state classification and structure–function dynamical modeling, while supporting real-time EEG/LFP-based monitoring. The proposed framework is validated for its efficacy in predicting brain states and elucidating connectivity patterns, thereby overcoming conventional black-box limitations. It significantly enhances the capacity to resolve the specificity of consciousness markers and establishes its core potential for translational applications in clinical neuroscience.

0 citationsRead paper

Language as an Independent Information Layer: A Conceptual Model of Communication, Cognition and Decision-Making

Sep 26, 2026

This study addresses the disconnect between statistical methods and semantic causal logic in enterprise knowledge bases, which impedes the precise identification of business bottlenecks. To bridge this gap, we propose a novel framework that conceptualizes language as an independent dynamic information layer. By integrating probabilistic vector spaces with ontological modeling and grounding the approach in dynamical systems theory, we construct a multi-dimensional solution space mapping framework that deeply couples statistical features with semantic logic. This methodology significantly enhances both knowledge extraction efficiency and its adaptability to business processes. Furthermore, it enables the accurate localization and effective resolution of logical bottlenecks within operational workflows. Ultimately, this work establishes a new paradigm for optimizing complex cognitive decision-making in enterprise environments.

0 citationsRead paper

Surv-IPTB: An Attention-Based Model for Estimating Individual Probability of Treatment Benefit with Survival Data

Aug 06, 2026

This study addresses the challenge of estimating individual probability of treatment benefit (IPTB) in survival analysis, particularly under complex scenarios involving heavy right-censoring and nonlinear treatment effects. The authors propose a novel approach that reformulates IPTB estimation as a pairwise binary classification problem between treated and control patients. By introducing a learnable query-key attention mechanism, the method flexibly aggregates cross-group comparison information and, for the first time, explicitly models censoring-induced uncertainty through interval-based probability representation, enabling end-to-end soft probabilistic learning. Extensive experiments on diverse nonlinear synthetic datasets demonstrate that the proposed method significantly outperforms established baselines—including T- and S-learners combined with random survival forests, Cox proportional hazards models, and Beran estimators—exhibiting remarkable robustness especially under high censoring rates and weak treatment effects.

0 citationsRead paper

Electronic Bursting Neuron: design, equations and hardware implementation

Jul 02, 2026

Existing electronic neuron designs struggle to simultaneously achieve low complexity, functional completeness, and mathematical tractability, thereby limiting their applicability in spiking neural networks. This work proposes a novel bursting electronic neuron architecture grounded in phase-locked loop system equations, employing a hybrid design paradigm that prioritizes target dynamics followed by hardware-aware reverse adaptation. By innovatively integrating phenomenological modeling with circuit simplification strategies, the approach circumvents both direct implementation of complex biophysical models and post-hoc equation fitting. The resulting neuron exhibits a compact structure, controllable dynamics, and strong theoretical analyzability while remaining amenable to hardware realization, making it well-suited for efficient modeling of individual neurons and small-scale neural circuits.

0 citationsRead paper

Attention-Based Estimation of the Individual Treatment Benefit Probability under Dose Variation

Jun 11, 2026

This study addresses the limitation of existing methods that are largely confined to binary treatment settings and struggle to accurately estimate individual probability of treatment benefit (IPTB) under discrete multi-dose interventions. The authors reformulate IPTB estimation as a binary classification problem targeting the sign of individual treatment effects. They construct pseudo-labels using pairs of covariate-similar samples and, for the first time, introduce an attention mechanism to aggregate information, replacing conventional Nadaraya-Watson kernel regression. This approach overcomes the binary treatment constraint and enables flexible modeling of personalized benefit probabilities across multiple doses. Experimental results on both real-world and synthetic data demonstrate that the proposed attention-based aggregation consistently outperforms kernel methods under various challenging conditions—including covariate shift, varying sample sizes, and heterogeneous treatment responses—providing a robust foundation for personalized dose decision-making.

0 citationsRead paper

Recent publications

Latest Papers

Deep Learning Methods in Neuroscience: From Modeling Molecular Mechanisms to Classifying States of Consciousness

Sep 28, 2026

This study addresses the limited model interpretability and lack of standardization in classifying states of consciousness and modeling anesthetized brain dynamics by proposing a hybrid, generalizable architecture that integrates physiological priors. By combining deep neural networks, multiscale computational modeling, and clustering algorithms, this work jointly analyzes EEG, fMRI, and LFP data to achieve automated brain state classification and structure–function dynamical modeling, while supporting real-time EEG/LFP-based monitoring. The proposed framework is validated for its efficacy in predicting brain states and elucidating connectivity patterns, thereby overcoming conventional black-box limitations. It significantly enhances the capacity to resolve the specificity of consciousness markers and establishes its core potential for translational applications in clinical neuroscience.

0 citationsRead paper

Language as an Independent Information Layer: A Conceptual Model of Communication, Cognition and Decision-Making

Sep 26, 2026

This study addresses the disconnect between statistical methods and semantic causal logic in enterprise knowledge bases, which impedes the precise identification of business bottlenecks. To bridge this gap, we propose a novel framework that conceptualizes language as an independent dynamic information layer. By integrating probabilistic vector spaces with ontological modeling and grounding the approach in dynamical systems theory, we construct a multi-dimensional solution space mapping framework that deeply couples statistical features with semantic logic. This methodology significantly enhances both knowledge extraction efficiency and its adaptability to business processes. Furthermore, it enables the accurate localization and effective resolution of logical bottlenecks within operational workflows. Ultimately, this work establishes a new paradigm for optimizing complex cognitive decision-making in enterprise environments.

0 citationsRead paper

Surv-IPTB: An Attention-Based Model for Estimating Individual Probability of Treatment Benefit with Survival Data

Aug 06, 2026

This study addresses the challenge of estimating individual probability of treatment benefit (IPTB) in survival analysis, particularly under complex scenarios involving heavy right-censoring and nonlinear treatment effects. The authors propose a novel approach that reformulates IPTB estimation as a pairwise binary classification problem between treated and control patients. By introducing a learnable query-key attention mechanism, the method flexibly aggregates cross-group comparison information and, for the first time, explicitly models censoring-induced uncertainty through interval-based probability representation, enabling end-to-end soft probabilistic learning. Extensive experiments on diverse nonlinear synthetic datasets demonstrate that the proposed method significantly outperforms established baselines—including T- and S-learners combined with random survival forests, Cox proportional hazards models, and Beran estimators—exhibiting remarkable robustness especially under high censoring rates and weak treatment effects.

0 citationsRead paper

Electronic Bursting Neuron: design, equations and hardware implementation

Jul 02, 2026

Existing electronic neuron designs struggle to simultaneously achieve low complexity, functional completeness, and mathematical tractability, thereby limiting their applicability in spiking neural networks. This work proposes a novel bursting electronic neuron architecture grounded in phase-locked loop system equations, employing a hybrid design paradigm that prioritizes target dynamics followed by hardware-aware reverse adaptation. By innovatively integrating phenomenological modeling with circuit simplification strategies, the approach circumvents both direct implementation of complex biophysical models and post-hoc equation fitting. The resulting neuron exhibits a compact structure, controllable dynamics, and strong theoretical analyzability while remaining amenable to hardware realization, making it well-suited for efficient modeling of individual neurons and small-scale neural circuits.

0 citationsRead paper

Attention-Based Estimation of the Individual Treatment Benefit Probability under Dose Variation

Jun 11, 2026

This study addresses the limitation of existing methods that are largely confined to binary treatment settings and struggle to accurately estimate individual probability of treatment benefit (IPTB) under discrete multi-dose interventions. The authors reformulate IPTB estimation as a binary classification problem targeting the sign of individual treatment effects. They construct pseudo-labels using pairs of covariate-similar samples and, for the first time, introduce an attention mechanism to aggregate information, replacing conventional Nadaraya-Watson kernel regression. This approach overcomes the binary treatment constraint and enables flexible modeling of personalized benefit probabilities across multiple doses. Experimental results on both real-world and synthetic data demonstrate that the proposed attention-based aggregation consistently outperforms kernel methods under various challenging conditions—including covariate shift, varying sample sizes, and heterogeneous treatment responses—providing a robust foundation for personalized dose decision-making.

0 citationsRead paper