arXiv subjects

Yijie Huang

Publications and source records attributed to Yijie Huang.

At least 19 recordsLinked to original sources

arXiv2026

HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes

Multimodal harmful meme detection is typically formulated as image--text harmfulness classification. A model may correctly predict harmfulness while misidentifying the attacked target or its supporting evidence. We therefore extend harmful meme detection with fine-grained target identification, asking what type of target is attacked, who is targeted, and where the target appears in the meme. The model predicts harmfulness for every meme and, for harmful memes, outputs the target category, target entity, textual mention, and visual region. To support this task, we introduce Meme3W, which unifies multiple public harmful meme datasets and provides human-verified annotations for harmful instances. We further introduce Joint Record Accuracy (JRA), a strict record-level metric requiring the harmfulness label and all target-identification fields to be jointly correct. Experiments with representative multimodal large language models reveal a substantial gap between harmfulness accuracy and JRA. To narrow this gap, we propose HarmTrace, an anchor-calibrated decoupled optimization framework. HarmTrace strengthens target-entity supervision through entity-aware supervised fine-tuning. It then applies Conditional Target-identification Policy Optimization (CTPO) to decouple harmfulness and target-identification advantages, restricting target-identification optimization to label-correct responses for harmful examples. CTPO uses a Virtual Positive Anchor (VPA) as a fully correct reference for target-identification advantage normalization. HarmTrace improves both JRA and harmfulness accuracy across the evaluated backbones, with JRA on the Qwen3-VL-8B backbone increasing from 17.58\% to 52.51\%. Our code is publicly available at https://github.com/llly1234/HarmTrace-for-Harmful-Memes.

cs.CV↗

arXiv2026

Continuous-time reinforcement learning for optimal switching over multiple regimes

This paper studies the continuous-time reinforcement learning (RL) for optimal switching problems across multiple regimes. We consider a type of exploratory formulation under entropy regularization where the agent randomizes both the timing of switches and the selection of regimes through the generator matrix of an associated continuous-time finite-state Markov chain. We establish the well-posedness of the associated system of Hamilton-Jacobi-Bellman (HJB) equations and provide a characterization of the optimal policy. The policy improvement and the convergence of the policy iterations are rigorously established by analyzing the system of equations. We also show that the value function in the exploratory formulation converges to the one in the classical formulation as the temperature parameter vanishes. Finally, a model-free reinforcement learning algorithm is devised and implemented by invoking the policy evaluation based on the martingale characterization. Our numerical examples with financial applications illustrate the effectiveness and efficiency of the proposed RL algorithm.

math.OC↗

arXiv2026

Dynamic Pricing for a Two-Sided Data Market Platform

We study a continuous-time dynamic pricing problem for a data platform that purchases raw data from privacy-sensitive providers and sells data products to consumers. The platform controls both the acquisition price offered to providers and the selling price charged to consumers. Provider and consumer arrivals are modeled by point processes whose intensities depend on the platform's current data stock, capturing feedback between data accumulation and market participation. We formulate the platform's problem as an infinite-horizon stochastic control problem with a jump-diffusion state process and derive the associated nonlinear integro-differential HJB equation. We prove that the value function is the unique viscosity solution, establish classical regularity under suitable conditions, and verify the optimal feedback pricing policy. Finally, we conduct numerical analyses to examine the influences of model parameters on the optimal pricing policies.

math.OC↗

arXiv2026

Optimal Consumption and Retirement Time under Shortfall Risk Measure

This paper studies the optimal portfolio, consumption, and endogenous early retirement problem within a benchmark tracking framework by incorporating a new relative performance evaluation. In this framework, the investor maximizes expected lifetime consumption utility while managing the maximum wealth shortfall relative to a benchmark, with shortfall-management costs that may differ before and after retirement. Mathematically, the problem is a hybrid stochastic control problem involving both regular controls and an optimal stopping time, in which the running maximum process records the investor's largest benchmark shortfall. We introduce an auxiliary reflected state process and establish an equivalent hybrid stochastic control problem. By proving the convex duality theorem, we technically transform the original problem into a two-dimensional pure optimal stopping problem with state reflection. This enables us to characterize the geometric structure of the stopping set and derive the feedback-form optimal retirement boundary, as well as optimal portfolio and consumption policies. Analytical examples and numerical simulations reveal a two-stage structure with more conservative investment and more aggressive consumption after retirement. Driven by the retirement option, the expected largest shortfall risk follows a pronounced U-shaped pattern with respect to wealth. Shortfall management costs, labor income, and leisure preference significantly influence retirement timing, investment, and consumption.

math.OC↗

arXiv2026

Deterministic Policy Gradient for Learning Equilibrium in Time-Inconsistent Control Problems

In this paper, we develop a continuous-time model-free reinforcement learning algorithm to learn deterministic equilibrium policies in general time-inconsistent control problems. Utilizing the extended Hamilton-Jacobi-Bellman system, we recast the original time-inconsistent problem into an equivalent two-stage problem. In the first stage, for given auxiliary functions, we employ the deterministic policy gradient approach to learn an optimal policy in an auxiliary time-consistent control problem. In the second stage, given the updated policy, we exploit the inner fixed point iterations and some martingale characterizations to learn the auxiliary functions. As a theoretical contribution, we provide some mild model assumptions and establish the convergence of inner fixed point iterations. By repeating this actor-critic style of iterations across two stages, our algorithm aims to learn the equilibrium under different sources of time-inconsistency in a unified manner. The superior effectiveness of the proposed algorithm are illustrated in two classical financial applications with time-inconsistency: mean-variance portfolio management and optimal tracking portfolio under non-exponential discounting.

q-fin.CP↗

arXiv2026

Transferable Graph Condensation from the Causal Perspective

The increasing scale of graph datasets has significantly improved the performance of graph representation learning methods, but it has also introduced substantial training challenges. Graph dataset condensation techniques have emerged to compress large datasets into smaller yet information-rich datasets, while maintaining similar test performance. However, these methods strictly require downstream applications to match the original dataset and task, which often fails in cross-task and cross-domain scenarios. To address these challenges, we propose a novel causal-invariance-based and transferable graph dataset condensation method, named TGCC, providing effective and transferable condensed datasets. Specifically, to preserve domain-invariant knowledge, we first extract domain causal-invariant features from the spatial domain of the graph using causal interventions. Then, to fully capture the structural and feature information of the original graph, we perform enhanced condensation operations. Finally, through spectral-domain enhanced contrastive learning, we inject the causal-invariant features into the condensed graph, ensuring that the compressed graph retains the causal information of the original graph. Experimental results on five public datasets and our novel FinReport dataset demonstrate that TGCC achieves up to a 13.41% improvement in cross-task and cross-domain complex scenarios compared to existing methods, and achieves state-of-the-art performance on 5 out of 6 datasets in the single dataset and task scenario.

cs.LG↗

arXiv2026

Mean Field Game of Optimal Tracking Portfolio

This paper studies the mean field game (MFG) problem arising from a large population competition in fund management, featuring a new type of relative performance via the benchmark tracking. In the $n$-player model, each agent aims to minimize the expected largest shortfall of the wealth with reference to the benchmark process, which is modeled by a linear combination of the population's average wealth process and a market index process. With a continuum of agents, we formulate the MFG problem with a reflected state process. We establish the existence of the mean field equilibrium (MFE) using the partial differential equation (PDE) approach. Firstly, by applying the dual transform, the best response control of the representative agent can be characterized in analytical form in terms of a dual reflected diffusion process. As a novel contribution, we verify the consistency condition of the MFE in separated domains with the help of the duality relationship and properties of the dual process. Moreover, based on the MFE, we construct an approximate Nash equilibrium for the $n$-player game when the number $n$ is sufficiently large.

math.OC↗

arXiv2026

Traceable Latent Variable Discovery Based on Multi-Agent Collaboration

Revealing the underlying causal mechanisms in the real world is crucial for scientific and technological progress. Despite notable advances in recent decades, the lack of high-quality data and the reliance of traditional causal discovery algorithms (TCDA) on the assumption of no latent confounders, as well as their tendency to overlook the precise semantics of latent variables, have long been major obstacles to the broader application of causal discovery. To address this issue, we propose a novel causal modeling framework, TLVD, which integrates the metadata-based reasoning capabilities of large language models (LLMs) with the data-driven modeling capabilities of TCDA for inferring latent variables and their semantics. Specifically, we first employ a data-driven approach to construct a causal graph that incorporates latent variables. Then, we employ multi-LLM collaboration for latent variable inference, modeling this process as a game with incomplete information and seeking its Bayesian Nash Equilibrium (BNE) to infer the possible specific latent variables. Finally, to validate the inferred latent variables across multiple real-world web-based data sources, we leverage LLMs for evidence exploration to ensure traceability. We comprehensively evaluate TLVD on three de-identified real patient datasets provided by a hospital and two benchmark datasets. Extensive experimental results confirm the effectiveness and reliability of TLVD, with average improvements of 32.67% in Acc, 62.21% in CAcc, and 26.72% in ECit across the five datasets.

cs.LG↗

arXiv2026

Continuous-time q-Learning for Jump-Diffusion Models under Tsallis Entropy

This paper studies the continuous-time reinforcement learning in jump-diffusion models by featuring the q-learning (the continuous-time counterpart of Q-learning) under Tsallis entropy regularization. Contrary to the Shannon entropy, the general form of Tsallis entropy renders the optimal policy not necessarily a Gibbs measure. Herein, the Lagrange multiplier and KKT condition are needed to ensure that the learned policy is a probability density function. As a consequence, the characterization of the optimal policy using the q-function also involves a Lagrange multiplier. In response, we establish the martingale characterization of the q-function and devise two q-learning algorithms depending on whether the Lagrange multiplier can be derived explicitly or not. In the latter case, we consider different parameterizations of the optimal q-function and the optimal policy, and update them alternatively in an Actor-Critic manner. We also study two numerical examples, namely, an optimal liquidation problem in dark pools and a non-LQ control problem. It is interesting to see therein that the optimal policies under the Tsallis entropy regularization can be characterized explicitly, which are distributions concentrated on some compact support. The satisfactory performance of our q-learning algorithms is illustrated in each example.

math.OC↗

arXiv2025

An extended Merton problem with relaxed benchmark tracking

This paper studies Merton's problem in an extended formulation by incorporating the benchmark tracking on the wealth process. We consider a tracking formulation where the fund manager aims to maximize the trade-off between the expected utility of consumption and the expected largest shortfall of the wealth with reference to the benchmark level. Equivalently, the problem can be interpreted as a mixed stochastic control problem if a fictitious capital injection singular control is allowed, subjecting to the dynamic constraint that the wealth process compensated by the costly capital injection outperforms the benchmark at all times. By considering an auxiliary state process, we formulate an equivalent stochastic control problem with state reflections at zero. For general utility functions and Ito's diffusion benchmark process, we develop a convex duality theorem, new to the literature, to the auxiliary stochastic control problem with state reflections in which the dual process also exhibits reflections from above. For CRRA utility and geometric Brownian motion benchmark process, we further derive the optimal portfolio and consumption in feedback form using the new duality theorem, allowing us to discuss some interesting financial implications induced by the additional risk-taking from the capital injection and the goal of tracking.

math.OC↗

arXiv2025

Affective Computing in the Era of Large Language Models: A Survey from the NLP Perspective

Affective Computing (AC) integrates computer science, psychology, and cognitive science to enable machines to recognize, interpret, and simulate human emotions across domains such as social media, finance, healthcare, and education. AC commonly centers on two task families: Affective Understanding (AU) and Affective Generation (AG). While fine-tuned pre-trained language models (PLMs) have achieved solid AU performance, they often generalize poorly across tasks and remain limited for AG, especially in producing diverse, emotionally appropriate responses. The advent of Large Language Models (LLMs) (e.g., ChatGPT and LLaMA) has catalyzed a paradigm shift by offering in-context learning, broader world knowledge, and stronger sequence generation. This survey presents an NLP-oriented overview of AC in the LLM era. We (i) consolidate traditional AC tasks and preliminary LLM-based studies; (ii) review adaptation techniques that improve AU/AG, including Instruction Tuning (full and parameter-efficient methods such as LoRA, P-/Prompt-Tuning), Prompt Engineering (zero/few-shot, chain-of-thought, agent-based prompting), and Reinforcement Learning. For the latter, we summarize RL from human preferences (RLHF), verifiable/programmatic rewards (RLVR), and AI feedback (RLAIF), which provide preference- or rule-grounded optimization signals that can help steer AU/AG toward empathy, safety, and planning, achieving finer-grained or multi-objective control. To assess progress, we compile benchmarks and evaluation practices for both AU and AG. We also discuss open challenges-from ethics, data quality, and safety to robust evaluation and resource efficiency-and outline research directions. We hope this survey clarifies the landscape and offers practical guidance for building affect-aware, reliable, and responsible LLM systems.

cs.CL↗

arXiv2025

Optimal consumption under relaxed benchmark tracking and consumption drawdown constraint

This paper studies an optimal consumption problem with both relaxed benchmark tracking and consumption drawdown constraint, leading to a stochastic control problem with dynamic state-control constraints. In our relaxed tracking formulation, it is assumed that the fund manager can strategically inject capital to the fund account such that the total capital process always outperforms the benchmark process, which is described by a geometric Brownian motion. We first transform the original regular-singular control problem with state-control constraints into an equivalent regular control problem with a reflected state process and consumption drawdown constraint. By utilizing the dual transform and the optimal consumption behavior, we then turn to study the linear dual PDE with both Neumann boundary condition and free boundary condition in a piecewise manner across different regions. Using the smoothfit principle and the super-contact condition, we derive the closed-form solution of the dual PDE, and obtain the optimal investment and consumption in feedback form. We then prove the verification theorem on optimality by some novel arguments with the aid of an auxiliary reflected dual process and some technical estimations. Some numerical examples and financial insights are also presented.

math.OC↗

arXiv2025

CLIP Brings Better Features to Visual Aesthetics Learners

Image Aesthetics Assessment (IAA) is a challenging task due to its subjective nature and expensive manual annotations. Recent large-scale vision-language models, such as Contrastive Language-Image Pre-training (CLIP), have shown their promising representation capability for various downstream tasks. However, the application of CLIP to resource-constrained and low-data IAA tasks remains limited. While few attempts to leverage CLIP in IAA have mainly focused on carefully designed prompts, we extend beyond this by allowing models from different domains and with different model sizes to acquire knowledge from CLIP. To achieve this, we propose a unified and flexible two-phase CLIP-based Semi-supervised Knowledge Distillation (CSKD) paradigm, aiming to learn a lightweight IAA model while leveraging CLIP's strong generalization capability. Specifically, CSKD employs a feature alignment strategy to facilitate the distillation of heterogeneous CLIP teacher and IAA student models, effectively transferring valuable features from pre-trained visual representations to two lightweight IAA models, respectively. To efficiently adapt to downstream IAA tasks in a low-data regime, the two strong visual aesthetics learners then conduct distillation with unlabeled examples for refining and transferring the task-specific knowledge collaboratively. Extensive experiments demonstrate that the proposed CSKD achieves state-of-the-art performance on multiple widely used IAA benchmarks. Furthermore, analysis of attention distance and entropy before and after feature alignment shows the effective transfer of CLIP's feature representation to IAA models, which not only provides valuable guidance for the model initialization of IAA but also enhances the aesthetic feature representation of IAA models. Code will be made publicly available.

cs.CV↗

arXiv2025

Optimal consumption under adjustment costs with respect to multiple reference levels

This paper studies a type of consumption preference where some adjustment costs are incured whenever the past spending maximum and the past spending minimum records are updated. This preference can capture the adverse effects of the historical consumption high and low values on the agent's consumption performance, thereby matching with some empirically observed smooth consumption patterns. By employing the dual transform, the smooth-fit conditions and the super-contact conditions, we obtain the closed-form solution of the dual PDE problem, and can characterize the optimal investment and consumption controls in the piecewise feedback form. We provide the rigorous proof of the verification theorem and compensate the theoretical findings with some numerical examples and financial implications.

math.OC↗

arXiv2024

A decomposition-homogenization method for Robin boundary problems on the nonnegative orthant

This paper studies the existence and uniqueness of a classical solution to a type of Robin boundary problems on the nonnegative orthant. We propose a new decomposition-homogenization method for the Robin boundary problem based on probabilistic representations, which leads to two auxiliary Robin boundary problems admitting some simplified probabilistic representations. The auxiliary probabilistic representations allow us to establish the existence of a unique classical solution to the original Robin boundary problem using some stochastic flow analysis.

math.PR↗

arXiv2024

On optimal tracking portfolio in incomplete markets: The reinforcement learning approach

This paper studies an infinite horizon optimal tracking portfolio problem using capital injection in incomplete market models. The benchmark process is modelled by a geometric Brownian motion with zero drift driven by some unhedgeable risk. The relaxed tracking formulation is adopted where the fund account compensated by the injected capital needs to outperform the benchmark process at any time, and the goal is to minimize the cost of the discounted total capital injection. When model parameters are known, we formulate the equivalent auxiliary control problem with reflected state dynamics, for which the classical solution of the HJB equation with Neumann boundary condition is obtained explicitly. When model parameters are unknown, we introduce the exploratory formulation for the auxiliary control problem with entropy regularization and develop the continuous-time q-learning algorithm in models of reflected diffusion processes. In some illustrative numerical example, we show the satisfactory performance of the q-learning algorithm.

q-fin.PM↗

arXiv2024

Minstrel: Structural Prompt Generation with Multi-Agents Coordination for Non-AI Experts

LLMs have demonstrated commendable performance across diverse domains. Nevertheless, formulating high-quality prompts to assist them in their work poses a challenge for non-AI experts. Existing research in prompt engineering suggests somewhat scattered optimization principles and designs empirically dependent prompt optimizers. Unfortunately, these endeavors lack a structural design, incurring high learning costs and it is not conducive to the iterative updating of prompts, especially for non-AI experts. Inspired by structured reusable programming languages, we propose LangGPT, a structural prompt design framework. Furthermore, we introduce Minstrel, a multi-generative agent system with reflection to automate the generation of structural prompts. Experiments and the case study illustrate that structural prompts generated by Minstrel or written manually significantly enhance the performance of LLMs. Furthermore, we analyze the ease of use of structural prompts through a user survey in our online community.

cs.CL↗

arXiv2024

LangGPT: Rethinking Structured Reusable Prompt Design Framework for LLMs from the Programming Language

LLMs have demonstrated commendable performance across diverse domains. Nevertheless, formulating high-quality prompts to instruct LLMs proficiently poses a challenge for non-AI experts. Existing research in prompt engineering suggests somewhat scattered optimization principles and designs empirically dependent prompt optimizers. Unfortunately, these endeavors lack a structured design template, incurring high learning costs and resulting in low reusability. In addition, it is not conducive to the iterative updating of prompts. Inspired by structured reusable programming languages, we propose LangGPT, a dual-layer prompt design framework as the programming language for LLMs. LangGPT has an easy-to-learn normative structure and provides an extended structure for migration and reuse. Experiments illustrate that LangGPT significantly enhances the performance of LLMs. Moreover, the case study shows that LangGPT leads LLMs to generate higher-quality responses. Furthermore, we analyzed the ease of use and reusability of LangGPT through a user survey in our online community.

cs.SE↗