Selected work

Representative Papers

VIANA: character Value-enhanced Intensity Assessment via domain-informed Neural Architecture

Apr 01, 2026

This study addresses the challenging problem of predicting the perceived intensity of odorant molecules by proposing a tripartite modeling framework that integrates molecular structure, olfactory semantics, and dose–response biological principles. The approach uniquely combines graph convolutional networks (GCNs) for molecular topology, semantic embeddings derived from the primary odor map (POM), and the Hill equation to model nonlinear response dynamics. To mitigate information redundancy in multi-source knowledge transfer, the authors introduce a PCA-based semantic signal distillation strategy. Evaluated on a test set, the model achieves exceptional performance with an R² of 0.996 and an MSE of 0.19, substantially outperforming existing baselines. It accurately captures key perceptual characteristics—including saturation ceilings, detection thresholds, and distinctive odor representations—thereby effectively bridging the gap between molecular structure and human olfactory perception.

0 citationsRead paper

NewPINNs: Physics-Informing Neural Networks Using Conventional Solvers for Partial Differential Equations

Jan 23, 2026

This work proposes NewPINNs, a novel framework that addresses the well-known optimization failures of traditional physics-informed neural networks (PINNs) when solving partial differential equations (PDEs), which often stem from ill-conditioned residual loss formulations, sensitivity to loss weighting, and challenges posed by stiffness or strong nonlinearity. NewPINNs uniquely integrates classical numerical solvers—such as finite volume, finite element, and spectral methods—directly into the neural network training process. Instead of relying on explicit PDE residual and boundary condition losses, the method enforces consistency between the neural network predictions and the states evolved by the embedded numerical solver through a pull-push interaction mechanism. This approach effectively circumvents common failure modes of PINNs, significantly enhancing solution stability and accuracy across a range of forward and inverse PDE problems, particularly in stiff and highly nonlinear regimes.

0 citationsRead paper

Prompted Policy Search: Reinforcement Learning through Linguistic and Numerical Reasoning in LLMs

Nov 26, 2025

Traditional reinforcement learning (RL) relies solely on scalar rewards, limiting its ability to incorporate rich semantic information—such as natural language instructions, commonsense knowledge, or domain-specific constraints. To address this, we propose Prompted Policy Search (ProPS), the first RL framework to directly integrate large language models (LLMs) into the core of policy optimization. ProPS jointly leverages natural language prompts—encoding task objectives, domain knowledge, and behavioral constraints—with numerical rewards to enable semantic-numerical co-reasoning. Crucially, it requires no LLM fine-tuning, instead harnessing in-context learning for policy updates and gradient-free numerical optimization. This design markedly improves exploration efficiency and sample efficiency. Evaluated on 15 Gymnasium benchmark tasks, ProPS outperforms mainstream algorithms—including PPO and SAC—on 8 tasks; performance gains are especially pronounced when domain knowledge is injected. ProPS advances interpretable, human-aligned, and general-purpose RL.

0 citationsRead paper

SEER: The Span-based Emotion Evidence Retrieval Benchmark

Oct 03, 2025

This paper introduces the span-level sentiment evidence detection task, which aims to precisely localize textual segments conveying sentiment—contrasting with conventional sentence-level sentiment classification—and thereby supports applications requiring fine-grained sentiment understanding, such as empathetic dialogue systems and clinical decision support. To this end, we construct the first manually annotated, multi-level benchmark (span-level labels for both single sentences and five-sentence paragraphs), systematically evaluating 14 open-source large language models. Our key contribution is reframing sentiment analysis from *discriminating sentiment categories* to *localizing sentiment-supporting evidence*, thereby advancing model interpretability and mechanistic understanding of sentiment expression. Experiments reveal that while certain models approach human performance on single-sentence detection, their accuracy degrades markedly on longer contexts, exposing critical limitations—including keyword dependency and high false-positive rates on neutral text. This benchmark establishes a novel evaluation paradigm and foundational infrastructure for explainable sentiment computation.

0 citationsRead paper

Accelerate Creation of Product Claims Using Generative AI

Sep 24, 2025

Product claim authoring is time-intensive yet critically influences consumer decision-making. This paper proposes an end-to-end automated framework integrating semantic retrieval, consumer tone matching, large language model (LLM) in-context learning and fine-tuning, and synthetic consumer simulation to enable efficient claim search, generation, optimization, and impact estimation. Our key contribution lies in jointly modeling consumer semantic preferences with controllable LLM generation, augmented by an interpretable synthetic user feedback loop that drives iterative refinement. Evaluated across multiple consumer goods enterprises, the framework reduces claim development cycle time by 62% on average, cuts authoring costs by 53%, and achieves a 92% business-validated quality acceptance rate. These results demonstrate both the efficacy and generalizability of our approach for marketing content generation.

0 citationsRead paper

Recent publications

Latest Papers

VIANA: character Value-enhanced Intensity Assessment via domain-informed Neural Architecture

Apr 01, 2026

This study addresses the challenging problem of predicting the perceived intensity of odorant molecules by proposing a tripartite modeling framework that integrates molecular structure, olfactory semantics, and dose–response biological principles. The approach uniquely combines graph convolutional networks (GCNs) for molecular topology, semantic embeddings derived from the primary odor map (POM), and the Hill equation to model nonlinear response dynamics. To mitigate information redundancy in multi-source knowledge transfer, the authors introduce a PCA-based semantic signal distillation strategy. Evaluated on a test set, the model achieves exceptional performance with an R² of 0.996 and an MSE of 0.19, substantially outperforming existing baselines. It accurately captures key perceptual characteristics—including saturation ceilings, detection thresholds, and distinctive odor representations—thereby effectively bridging the gap between molecular structure and human olfactory perception.

0 citationsRead paper

NewPINNs: Physics-Informing Neural Networks Using Conventional Solvers for Partial Differential Equations

Jan 23, 2026

This work proposes NewPINNs, a novel framework that addresses the well-known optimization failures of traditional physics-informed neural networks (PINNs) when solving partial differential equations (PDEs), which often stem from ill-conditioned residual loss formulations, sensitivity to loss weighting, and challenges posed by stiffness or strong nonlinearity. NewPINNs uniquely integrates classical numerical solvers—such as finite volume, finite element, and spectral methods—directly into the neural network training process. Instead of relying on explicit PDE residual and boundary condition losses, the method enforces consistency between the neural network predictions and the states evolved by the embedded numerical solver through a pull-push interaction mechanism. This approach effectively circumvents common failure modes of PINNs, significantly enhancing solution stability and accuracy across a range of forward and inverse PDE problems, particularly in stiff and highly nonlinear regimes.

0 citationsRead paper

Prompted Policy Search: Reinforcement Learning through Linguistic and Numerical Reasoning in LLMs

Nov 26, 2025

Traditional reinforcement learning (RL) relies solely on scalar rewards, limiting its ability to incorporate rich semantic information—such as natural language instructions, commonsense knowledge, or domain-specific constraints. To address this, we propose Prompted Policy Search (ProPS), the first RL framework to directly integrate large language models (LLMs) into the core of policy optimization. ProPS jointly leverages natural language prompts—encoding task objectives, domain knowledge, and behavioral constraints—with numerical rewards to enable semantic-numerical co-reasoning. Crucially, it requires no LLM fine-tuning, instead harnessing in-context learning for policy updates and gradient-free numerical optimization. This design markedly improves exploration efficiency and sample efficiency. Evaluated on 15 Gymnasium benchmark tasks, ProPS outperforms mainstream algorithms—including PPO and SAC—on 8 tasks; performance gains are especially pronounced when domain knowledge is injected. ProPS advances interpretable, human-aligned, and general-purpose RL.

0 citationsRead paper

SEER: The Span-based Emotion Evidence Retrieval Benchmark

Oct 03, 2025

This paper introduces the span-level sentiment evidence detection task, which aims to precisely localize textual segments conveying sentiment—contrasting with conventional sentence-level sentiment classification—and thereby supports applications requiring fine-grained sentiment understanding, such as empathetic dialogue systems and clinical decision support. To this end, we construct the first manually annotated, multi-level benchmark (span-level labels for both single sentences and five-sentence paragraphs), systematically evaluating 14 open-source large language models. Our key contribution is reframing sentiment analysis from *discriminating sentiment categories* to *localizing sentiment-supporting evidence*, thereby advancing model interpretability and mechanistic understanding of sentiment expression. Experiments reveal that while certain models approach human performance on single-sentence detection, their accuracy degrades markedly on longer contexts, exposing critical limitations—including keyword dependency and high false-positive rates on neutral text. This benchmark establishes a novel evaluation paradigm and foundational infrastructure for explainable sentiment computation.

0 citationsRead paper

Accelerate Creation of Product Claims Using Generative AI

Sep 24, 2025

Product claim authoring is time-intensive yet critically influences consumer decision-making. This paper proposes an end-to-end automated framework integrating semantic retrieval, consumer tone matching, large language model (LLM) in-context learning and fine-tuning, and synthetic consumer simulation to enable efficient claim search, generation, optimization, and impact estimation. Our key contribution lies in jointly modeling consumer semantic preferences with controllable LLM generation, augmented by an interpretable synthetic user feedback loop that drives iterative refinement. Evaluated across multiple consumer goods enterprises, the framework reduces claim development cycle time by 62% on average, cuts authoring costs by 53%, and achieves a 92% business-validated quality acceptance rate. These results demonstrate both the efficacy and generalizability of our approach for marketing content generation.

0 citationsRead paper