arXiv subjects

Chang Yang

Publications and source records attributed to Chang Yang.

At least 19 recordsLinked to original sources

arXiv2026

An Asymptotic-Preserving Micro--Macro Scheme for Plasma Simulations in Quasi-Neutral and Low-Mach-Number Regimes with Kinetic Upgrades

We propose an asymptotic-preserving micro--macro method bridging a kinetic description of electrons and a low-frequency reduced model in which the electrons are a massless, quasi-neutral fluid obeying the Boltzmann relation. Two features distinguish the construction. First, the fluid and low-Mach limits are coupled, so that the low-Mach stiffness is handled on a macroscopic system, where implicit treatment is affordable, rather than on the kinetic equations. Second, an auxiliary variable rescales the stiff force balance, turning the singular low-Mach limit into a regular limit of the augmented system, which is shown to remain non-degenerate uniformly in the Debye length as well. This matters at the discrete level: with an iterative linear solver, the stiffness induced by the small Mach number amplifies the solver residual, so that a scheme designed to be asymptotic-preserving in its time discretization alone loses that property once the full solution chain is taken into account. The proposed scheme retains it, with no tightening of the solver tolerance as the Mach number vanishes, and admits a post-processing variant that decouples the auxiliary variable and reduces the size of the linear system. Numerical experiments spanning distinct parameter regimes confirm the analysis: standard semi-implicit schemes lose low-Mach-number equilibrium under residual amplification, whereas the proposed schemes preserve it down to round-off.

math.NA↗

arXiv2026

The Agent Use of Agent Beings: Agent Cybernetics Is the Missing Science of Foundation Agents

LLM-based foundation agents that perceive, reason, and act across thousands of reasoning steps are rapidly becoming the dominant paradigm for deploying artificial intelligence in open-ended, long-horizon complex tasks. Despite this significance, the field remains overwhelmingly engineering-driven. Engineering practice has converged on useful primitives (tool loops, memory banks, harnesses, reflection steps), yet these are assembled by empirical trial and error rather than from first principles. Fundamental questions remain open: under what conditions does a long-running agent remain on-task? How should an agent respond when its environment exceeds its representational capacity? What architectural properties are necessary for safe self-improvement? We argue that cybernetics, the mid-twentieth-century science of control and communication in complex systems, provides the missing theoretical scaffold for foundation agents. By mapping six canonical laws of classical cybernetics onto six agent design principles, and synthesizing those principles into three engineering desiderata (reliability, lifelong running, and self-Improvement), we arrive at a framework termed Agent Cybernetics. Three application domains, code generation, computer use and automated research, exemplify the analytical framework of agent cybernetics by identifying failure modes and concrete engineering recommendations. We hope that agent cybernetics opens a new research venue and establishes the scientific foundation that foundation agents need for principled, reliable real-world deployment.

cs.AI↗

arXiv2026

Towards Brain MRI Foundation Models for the Clinic: Findings from the FOMO25 Challenge

Clinical deployment of automated brain MRI analysis faces a fundamental challenge: clinical data is heterogeneous and noisy, and high-quality labels are prohibitively costly to obtain. Self-supervised learning (SSL) can address this by leveraging the vast amounts of unlabeled data produced in clinical workflows to train robust \textit{foundation models} that adapt out-of-domain with minimal supervision. However, the development of foundation models for brain MRI has been limited by small pretraining datasets and in-domain benchmarking focused on high-quality, research-grade data. To address this gap, we organized the FOMO25 challenge as a satellite event at MICCAI 2025. FOMO25 provided participants with a large pretraining dataset, FOMO60K, and evaluated models on data sourced directly from clinical workflows in few-shot and out-of-domain settings. Tasks covered infarct classification, meningioma segmentation, and brain age regression, and considered both models trained on FOMO60K (method track) and any data (open track). Nineteen foundation models from sixteen teams were evaluated using a standardized containerized pipeline. Results show that (a) self-supervised pretraining improves generalization on clinical data under domain shift, with the strongest models trained \textit{out-of-domain} surpassing supervised baselines trained \textit{in-domain}. (b) No single pretraining objective benefits all tasks: MAE favors segmentation, hybrid reconstruction-contrastive objectives favor classification, and (c) strong performance was achieved by small pretrained models, and improvements from scaling model size and training duration did not yield reliable benefits.

cs.CV↗

arXiv2026

Design Once, Deploy at Scale: Template-Driven ML Development for Large Model Ecosystems

Modern computational advertising platforms typically rely on recommendation systems to predict user responses, such as click-through rates, conversion rates, and other optimization events. To support a wide variety of product surfaces and advertiser goals, these platforms frequently maintain an extensive ecosystem of machine learning (ML) models. However, operating at this scale creates significant development and efficiency challenges. Substantial engineering effort is required to regularly refresh ML models and propagate new techniques, which results in long latencies when deploying ML innovations across the ecosystem. We present a large-scale empirical study comparing model performance, efficiency, and ML technique propagation between a standardized model-building approach and independent per-model optimization in recommendation systems. To facilitate this standardization, we propose the Standard Model Template (SMT) -- a framework that generates high-performance models adaptable to diverse data distributions and optimization events. By utilizing standardized, composable ML model components, SMT reduces technique propagation complexity from $O(n \cdot 2^k)$ to $O(n + k)$ where $n$ is the number of models and $k$ the number of techniques. Evaluating an extensive suite of models over four global development cycles within Meta's production ads ranking ecosystem, our results demonstrate: (1) a 0.63% average improvement in cross-entropy at neutral serving capacity, (2) a 92% reduction in per-model iteration engineering time, and (3) a $6.3\times$ increase in technique-model pair adoption throughput. These findings challenge the conventional wisdom that diverse optimization goals inherently require diversified ML model design.

cs.AI↗

arXiv2026

Graph-based Agent Memory: Taxonomy, Techniques, and Applications

Memory emerges as the core module in the Large Language Model (LLM)-based agents for long-horizon complex tasks (e.g., multi-turn dialogue, game playing, scientific discovery), where memory can enable knowledge accumulation, iterative reasoning and self-evolution. Among diverse paradigms, graph stands out as a powerful structure for agent memory due to the intrinsic capabilities to model relational dependencies, organize hierarchical information, and support efficient retrieval. This survey presents a comprehensive review of agent memory from the graph-based perspective. First, we introduce a taxonomy of agent memory, including short-term vs. long-term memory, knowledge vs. experience memory, non-structural vs. structural memory, with an implementation view of graph-based memory. Second, according to the life cycle of agent memory, we systematically analyze the key techniques in graph-based agent memory, covering memory extraction for transforming the data into the contents, storage for organizing the data efficiently, retrieval for retrieving the relevant contents from memory to support reasoning, and evolution for updating the contents in the memory. Third, we summarize the open-sourced libraries and benchmarks that support the development and evaluation of self-evolving agent memory. We also explore diverse application scenarios. Finally, we identify critical challenges and future research directions. This survey aims to offer actionable insights to advance the development of more efficient and reliable graph-based agent memory systems. All the related resources, including research papers, open-source data, and projects, are collected for the community in https://github.com/DEEP-PolyU/Awesome-GraphMemory.

cs.AI↗

arXiv2025

Large Emotional World Model

The world is governed by both physical laws and affective dynamics. Physical laws govern state transitions, while affective dynamics shape human actions, decisions, and interactions. A world model that learns only physical laws can approximate the physical world, but not the human world. In this paper, we introduce human emotion as a key state variable in world models, enabling them to capture both future state transitions and their emotional causes. We first construct Emotion-Why-How (EWH), the first world model dataset centered on emotional state transitions, containing 10,850 emotion-aware transition tuples. Each tuple encodes the pre-state, pre-emotion, action, post-emotion, and post-state, supporting reasoning about why actions occur and how emotions reshape future states. Based on EWH, we propose the Large Emotional World Model (LEWM), which factorizes future prediction into two coupled steps: first predicting the future emotional state from the current context, and then conditioning future world-state prediction on the predicted emotion. Experiments show that LEWM brings consistent gains across world-state prediction, emotion understanding, and general reasoning tasks. It achieves up to 45.72% accuracy improvement on EWH, 3.94% on WorldNet, 17.47% F1 improvement on MELD, and a 6.10% gain on specific MMLU categories. These results demonstrate that incorporating emotion into world models enables more realistic simulation of human-centered environments and expands the predictive understanding of intelligent agents.

cs.CL↗

arXiv2025

Augmenting Intra-Modal Understanding in MLLMs for Robust Multimodal Keyphrase Generation

Multimodal keyphrase generation (MKP) aims to extract a concise set of keyphrases that capture the essential meaning of paired image-text inputs, enabling structured understanding, indexing, and retrieval of multimedia data across the web and social platforms. Success in this task demands effectively bridging the semantic gap between heterogeneous modalities. While multimodal large language models (MLLMs) achieve superior cross-modal understanding by leveraging massive pretraining on image-text corpora, we observe that they often struggle with modality bias and fine-grained intra-modal feature extraction. This oversight leads to a lack of robustness in real-world scenarios where multimedia data is noisy, along with incomplete or misaligned modalities. To address this problem, we propose AimKP, a novel framework that explicitly reinforces intra-modal semantic learning in MLLMs while preserving cross-modal alignment. AimKP incorporates two core innovations: (i) Progressive Modality Masking, which forces fine-grained feature extraction from corrupted inputs by progressively masking modality information during training; (ii) Gradient-based Filtering, that identifies and discards noisy samples, preventing them from corrupting the model's core cross-modal learning. Extensive experiments validate AimKP's effectiveness in multimodal keyphrase generation and its robustness across different scenarios.

cs.MM↗

arXiv2025

A Conservative Cascade Semi-Lagrangian Method for Solving the Vlasov Equation

The cascade remapping method, originally proposed by Nair et al. (2002) for atmospheric modeling, enables efficient and mass conservative semi Lagrangian (SL) transport through successive one dimensional remapping. While widely used in geophysical flows, its application to plasma kinetics remains limited. To exploit its potential advantages in conservation and scalability, this work applies the conservative cascade semi Lagrangian (CCSL) scheme to the Vlasov equation and related plasma models. A consistency analysis shows that the scheme attains second order spatial accuracy, with the dominant error arising from the geometric approximation of the backtracked region. Moreover, two improvements are introduced: a freestream preserving correction that ensures exact volume conservation, and a maximum principle limiter that suppresses spurious oscillations while maintaining positivity and mass conservation. Numerical tests, including linear advection, guiding center, and relativistic Vlasov Maxwell models, confirm the high accuracy, robustness, and long term stability of the improved CCSL method. Compared with the conservative semi Lagrangian (CSL) and the backward semi Lagrangian (BSL) schemes, it better preserves physical invariants under divergence free conditions, providing a robust and efficient framework for high-fidelity plasma kinetic simulations with good parallel scalability.

math.NA↗

arXiv2025

New measurement of $^{51}$V($\gamma$,1n) cross section through the refined monochromatic cross section extraction method

The Giant Dipole Resonance (GDR) in $^{51}$V has been a long-term conflicting interpretation, with existing photoneutron cross section data suggesting either a single peak or a pronounced splitting, leading to opposite conclusions on nuclear deformation. A new measurement of the $^{51}$V($\gamma$,1n) cross section, performed at the Shanghai Laser Electron Gamma Source (SLEGS) facility, employs a refined monochromatic cross section extraction method. By integrating Polynomial Regression and Support Vector Regression (SVR) for robust interpolation and extrapolation, the new extracted monoenergetic cross sections exhibit a single, broad peak with no evidence of GDR splitting. This result provides new support for a spherical or near-spherical shape of $^{51}$V. Furthermore, we found that deliberately overfitting the data using an SVR model reproduces multi-peak structures similar to those reported in historical datasets, implying that the previously claimed splitting might originated from analysis artifacts rather than physical phenomena.

nucl-ex↗

arXiv2025

Towards Frequency-Adaptive Learning for SAR Despeckling

Synthetic Aperture Radar (SAR) images are inherently corrupted by speckle noise, limiting their utility in high-precision applications. While deep learning methods have shown promise in SAR despeckling, most methods employ a single unified network to process the entire image, failing to account for the distinct speckle statistics associated with different spatial physical characteristics. It often leads to artifacts, blurred edges, and texture distortion. To address these issues, we propose SAR-FAH, a frequency-adaptive heterogeneous despeckling model based on a divide-and-conquer architecture. First, wavelet decomposition is used to separate the image into frequency sub-bands carrying different intrinsic characteristics. Inspired by their differing noise characteristics, we design specialized sub-networks for different frequency components. The tailored approach leverages statistical variations across frequencies, improving edge and texture preservation while suppressing noise. Specifically, for the low-frequency part, denoising is formulated as a continuous dynamic system via neural ordinary differential equations, ensuring structural fidelity and sufficient smoothness that prevents artifacts. For high-frequency sub-bands rich in edges and textures, we introduce an enhanced U-Net with deformable convolutions for noise suppression and enhanced features. Extensive experiments on synthetic and real SAR images validate the superior performance of the proposed model in noise removal and structural preservation.

cs.CV↗

arXiv2025

Hierarchical Dual-Head Model for Suicide Risk Assessment via MentalRoBERTa

Social media platforms have become important sources for identifying suicide risk, but automated detection systems face multiple challenges including severe class imbalance, temporal complexity in posting patterns, and the dual nature of risk levels as both ordinal and categorical. This paper proposes a hierarchical dual-head neural network based on MentalRoBERTa for suicide risk classification into four levels: indicator, ideation, behavior, and attempt. The model employs two complementary prediction heads operating on a shared sequence representation: a CORAL (Consistent Rank Logits) head that preserves ordinal relationships between risk levels, and a standard classification head that enables flexible categorical distinctions. A 3-layer Transformer encoder with 8-head multi-head attention models temporal dependencies across post sequences, while explicit time interval embeddings capture posting behavior dynamics. The model is trained with a combined loss function (0.5 CORAL + 0.3 Cross-Entropy + 0.2 Focal Loss) that simultaneously addresses ordinal structure preservation, overconfidence reduction, and class imbalance. To improve computational efficiency, we freeze the first 6 layers (50%) of MentalRoBERTa and employ mixed-precision training. The model is evaluated using 5-fold stratified cross-validation with macro F1 score as the primary metric.

cs.LG↗

arXiv2025

Macro-Level Correlational Analysis of Mental Disorders: Economy, Education, Society, and Technology Development

This paper quantifies the age-stratified global burden of four mental disorders in 27 regions from 1990 to 2021 using GBD 2021. To put it in detail, it links the age-standardized years of disability adjustment with 18 world development indicators across economic, educational, social and information technology sectors. Then, by means of Pearson correlation, mutual information, Granger causality and maximum information coefficient and other methods, the linear, nonlinear and lagged dependency relationships were evaluated. After research, it was found that there is a very prominent spatio-temporal heterogeneity among young people aged 20 to 39, and the coupling relationship is stronger. From the overall situation, education corresponds to a low burden. Unemployment corresponds to a high burden. Through lag analysis, it can be known that the influence time of economic and technological factors is relatively short, while that of educational factors is relatively long. These results highlight the macro determinants that play a role at different time scales and also provide population-level references for verifying computational mental health models and for intervention measures in specific regions and for specific ages.

stat.AP↗

arXiv2025

Intertwining periods, L-functions and local-global principles for distinction of automorphic representations

We provide a criterion for non-vanishing of period integrals on automorphic representations of a general linear group over a division algebra. We consider three different periods: linear periods, twisted-linear periods and Galois periods. Our criterion is a local-global principle, which is stated in terms of local distinction, a further local obstruction, and poles of certain global L-functions associated to the underlying involution via the Jacquet-Langlands correspondence. Our local-global principle follows from a new method, relying on the Maass-Selberg relations and a careful analysis of singularities of local and global intertwining periods. Our results generalize to inner forms, known results for split general linear groups. Moreover, our result for twisted linear periods is new even in the split situation. As a consequence of our local-global principle, we complete the proof of one direction of the Guo-Jacquet conjecture.

math.NT↗

arXiv2025

LAG: Logic-Augmented Generation from a Cartesian Perspective

Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet exhibit critical limitations in knowledge-intensive tasks, often generating hallucinations when faced with questions requiring specialized expertise. While retrieval-augmented generation (RAG) mitigates this by integrating external knowledge, it struggles with complex reasoning scenarios due to its reliance on direct semantic retrieval and lack of structured logical organization. Inspired by Cartesian principles from \textit{Discours de la m\'ethode}, this paper introduces Logic-Augmented Generation (LAG), a novel paradigm that reframes knowledge augmentation through systematic question decomposition, atomic memory bank and logic-aware reasoning. Specifically, LAG first decomposes complex questions into atomic sub-questions ordered by logical dependencies. It then resolves these sequentially, using prior answers to guide context retrieval for subsequent sub-questions, ensuring stepwise grounding in the logical chain. Experiments on four benchmarks demonstrate that LAG significantly improves accuracy and reduces hallucination over existing methods.

cs.CL↗

arXiv2025

Charmed baryon decays: $SU(3)_F$ breaking and CP violation

We present a comprehensive study of anti-triplet charmed baryon decays into octet baryons and pseudoscalar mesons using the $SU(3)_F$ flavor symmetry framework. By decomposing the flavor structure, all decay amplitudes are parametrized with a minimal set of irreducible amplitudes, and the K\"{o}rner-Pati-Woo theorem further reduces the number of independent parameters from the original 35 to 19 under exact symmetry when small terms proportional to $\lambda_b = V_{cb}^*V_{ub}$ are neglected. The independent number becomes 27 with leading $SU(3)_F$ breaking effects. A global fit to 51 experimental measurements yields a $\chi^2$/d.o.f. of 2.36 and provides precise values for the decay amplitudes. Notable discrepancies are observed in the fitted branching fractions of $\Xi_c^0 \to \Xi^- \pi^+$ and $\Xi_c^+ \to \Xi^- \pi^+ \pi^+$, which exceed current measurements by more than $2\sigma$. Incorporating final-state rescattering effects to recover the main effects of terms proportional to $\lambda_b$, we predict enhanced CP violation, with $A_{CP}$ reaching up to $10^{-3}$ in golden channels such as $\Xi_c^0 \to p K^-$ and $\Xi_c^0 \to \Sigma^+ \pi^-$. Our analysis also finds that the branching ratio for $\Xi_c^+ \to \Lambda \pi^+$ is enhanced to $(9.7 \pm 1.2) \times 10^{-4}$ due to significant cancellations.

hep-ph↗

arXiv2025

Resolving Latency and Inventory Risk in Market Making with Reinforcement Learning

The latency of the exchanges in Market Making (MM) is inevitable due to hardware limitations, system processing times, delays in receiving data from exchanges, the time required for order transmission to reach the market, etc. Existing reinforcement learning (RL) methods for Market Making (MM) overlook the impact of these latency, which can lead to unintended order cancellations due to price discrepancies between decision and execution times and result in undesired inventory accumulation, exposing MM traders to increased market risk. Therefore, these methods cannot be applied in real MM scenarios. To address these issues, we first build a realistic MM environment with random delays of 30-100 milliseconds for order placement and market information reception, and implement a batch matching mechanism that collects orders within every 500 milliseconds before matching them all at once, simulating the batch auction mechanisms adopted by some exchanges. Then, we propose Relaver, an RL-based method for MM to tackle the latency and inventory risk issues. The three main contributions of Relaver are: i) we introduce an augmented state-action space that incorporates order hold time alongside price and volume, enabling Relaver to optimize execution strategies under latency constraints and time-priority matching mechanisms, ii) we leverage dynamic programming (DP) to guide the exploration of RL training for better policies, iii) we train a market trend predictor, which can guide the agent to intelligently adjust the inventory to reduce the risk. Extensive experiments and ablation studies on four real-world datasets demonstrate that \textsc{Relaver} significantly improves the performance of state-of-the-art RL-based MM strategies across multiple metrics.

cs.LG↗

arXiv2025

FinMaster: A Holistic Benchmark for Mastering Full-Pipeline Financial Workflows with LLMs

Financial tasks are pivotal to global economic stability; however, their execution faces challenges including labor intensive processes, low error tolerance, data fragmentation, and tool limitations. Although large language models (LLMs) have succeeded in various natural language processing tasks and have shown potential in automating workflows through reasoning and contextual understanding, current benchmarks for evaluating LLMs in finance lack sufficient domain-specific data, have simplistic task design, and incomplete evaluation frameworks. To address these gaps, this article presents FinMaster, a comprehensive financial benchmark designed to systematically assess the capabilities of LLM in financial literacy, accounting, auditing, and consulting. Specifically, FinMaster comprises three main modules: i) FinSim, which builds simulators that generate synthetic, privacy-compliant financial data for companies to replicate market dynamics; ii) FinSuite, which provides tasks in core financial domains, spanning 183 tasks of various types and difficulty levels; and iii) FinEval, which develops a unified interface for evaluation. Extensive experiments over state-of-the-art LLMs reveal critical capability gaps in financial reasoning, with accuracy dropping from over 90% on basic tasks to merely 40% on complex scenarios requiring multi-step reasoning. This degradation exhibits the propagation of computational errors, where single-metric calculations initially demonstrating 58% accuracy decreased to 37% in multimetric scenarios. To the best of our knowledge, FinMaster is the first benchmark that covers full-pipeline financial workflows with challenging tasks. We hope that FinMaster can bridge the gap between research and industry practitioners, driving the adoption of LLMs in real-world financial practices to enhance efficiency and accuracy.

cs.AI↗

arXiv2025

Why Regression? Binary Encoding Classification Brings Confidence to Stock Market Index Price Prediction

Stock market indices serve as fundamental market measurement that quantify systematic market dynamics. However, accurate index price prediction remains challenging, primarily because existing approaches treat indices as isolated time series and frame the prediction as a simple regression task. These methods fail to capture indices' inherent nature as aggregations of constituent stocks with complex, time-varying interdependencies. To address these limitations, we propose Cubic, a novel end-to-end framework that explicitly models the adaptive fusion of constituent stocks for index price prediction. Our main contributions are threefold. i) Fusion in the latent space: we introduce the fusion mechanism over the latent embedding of the stocks to extract the information from the vast number of stocks. ii) Binary encoding classification: since regression tasks are challenging due to continuous value estimation, we reformulate the regression into the classification task, where the target value is converted to binary and we optimize the prediction of the value of each digit with cross-entropy loss. iii) Confidence-guided prediction and trading: we introduce the regularization loss to address market prediction uncertainty for the index prediction and design the rule-based trading policies based on the confidence. Extensive experiments across multiple stock markets and indices demonstrate that Cubic consistently outperforms state-of-the-art baselines in stock index prediction tasks, achieving superior performance on both forecasting accuracy metrics and downstream trading profitability.

q-fin.ST↗