arXiv subjects

Dan Liu

Publications and source records attributed to Dan Liu.

At least 19 recordsLinked to original sources

arXiv2026

Cost-Utility Alignment in LLM Agent Trajectories:Profiling,Attribution,Diagnosis,Adaptation,and Evaluation

LLM agents execute tasks through multi-step trajectories that accumulate cost in tokens, latency, monetary fees, and environmental risk while producing utility only at the aggregate task level. Prior surveys address inference optimization, agent capabilities, or evaluation in isolation, leaving practitioners without principled tools to determine whether a trajectory's resource expenditure is justified by its task contribution. We address this gap by developing a trajectory-centric cost-utility alignment framework that treats resource consumption and task contribution as dual ledgers over the same execution, organized around five analytical stages: cost profiling, utility attribution, misalignment diagnosis, targeted adaptation, and evaluation. Utility attribution is central to this structure: rather than relying on aggregate outcomes, it organizes contribution methods by evidential strength, from process proxies and information dependency to counterfactual replay, supplying the causal evidence that grounds diagnosis and guides adaptation. Using this framework, we analyze recent agent systems, attribution methods, and evaluation protocols covering efficiency, reliability, and economic value, as well as five forms of misalignment spanning cognitive and context use, external interaction, recovery-loop control, resource-capability allocation, and multi-agent coordination, together with their targeted adaptations. The result is a closed analytical loop connecting the cost side of agent execution to its utility side, providing a structured basis for resource-aware agent design and deployment.

cs.SE↗

arXiv2026

Generating Benchmark Health Data Using a Tabular Diffusion Transformer

Cross-Tabular Data Generation (CTDG) seeks to learn a generative model from multiple heterogeneous tables and produce new synthetic tabular datasets. However, existing synthetic tabular data generation methods are largely restricted to single-input-table scenarios and struggle to effectively handle multiple heterogeneous tables with diverse feature sets. To address this limitation, we propose a two-stage framework for cross-tabular data generation. In the first stage, each heterogeneous raw table is transformed into a standardized statistical table with the same set of columns across all tables. Each statistical table captures the marginal distributions of the original columns and the pairwise correlations among them. In the second stage, a diffusion transformer model is trained to capture structural patterns across these homogeneous statistical tables and to generate synthetic statistical tables. Synthetic raw tables are subsequently reconstructed from the generated statistical tables via multivariate Gaussian sampling followed by an inverse probability integral transform. This two-stage CTDG framework enables the learning of a unified generative model from multiple heterogeneous tables and supports the generation of an unlimited number of realistic synthetic heterogeneous tables. Experimental results demonstrate high fidelity in the learned statistical representations and a favorable fidelity-diversity trade-off in the generated synthetic data, validating the effectiveness of the proposed approach.

cs.LG↗

arXiv2026

CatalogAgent: A Supervisor-mediated Self-Learning System Enabling Context Engineering for GenAI Models

Product catalogs are the backbone of e-commerce sites, yet a large number of structured attributes (SAs) -- such as material, color, and shape -- often have missing values. Typically, SA values are extracted from product information, including titles and descriptions. While LLM-based generator-evaluator frameworks have demonstrated effectiveness for SA prediction -- where an LLM generates SA values and another evaluates them -- they face challenges when the Generator and Evaluator produce conflicting outputs, as either component can make mistakes. We introduce \texttt{CatalogAgent}, a novel agentic system that continuously improves Generator and Evaluator models for e-commerce catalog enrichment. When disagreements arise from (1) internal conflicts between the LLM-based Generator and Evaluator, or (2) external feedback from sellers on LLM outputs, a Supervisor Agent intervenes to mediate these conflicts and make final decisions. The system also incorporates a Memory Base and a Memory Summarizer that stores Supervisor Agent activities from individual cases and aggregates patterns into learnings. These learnings are fed back to the worker Generator and Evaluator LLMs, enabling self-improvement without human intervention. Through context engineering -- injecting learnings and insights into worker LLMs' contexts -- the system successfully transfers the Supervisor's capabilities to the Generator and Evaluator, improving their performance by 15.24\% and 13.98\%, respectively. Our experiments demonstrate a new paradigm of Supervisor Agent-mediated self-learning systems for improving generative AI model accuracy.

cs.AI↗

arXiv2026

PGUDA: Pressure-Guided Unsupervised Domain Adaptation with Cross-Modal Knowledge Distillation for sEMG-Based Gesture Recognition

Surface electromyography (sEMG)-based gesture recognition has emerged as a promising technology for natural human-computer interaction. However, its practical deployment remains challenging due to severe performance degradation caused by feature distribution discrepancies across different subjects and recording sessions. Although domain adaptation (DA) techniques are commonly employed to mitigate such discrepancies, conventional methods often struggle to effectively aligning sEMG features, primarily due to their inherent stochasticity and the scarcity of labeled data. To address these limitations, this paper proposes a novel Pressure-Guided Unsupervised Domain Adaptation (PGUDA) framework, which leverages the robustness and stability of pressure signals to introduce a cross-modal knowledge distillation strategy that transfers consistent physical semantics across modalities. Specifically, a teacher network trained on pressure signals guides an sEMG student network on unlabeled target domains, thereby regularizing the representation learning process with transferable and modality-invariant knowledge. Extensive experiments conducted on a self-collected multimodal dataset involving eleven subjects validate the effectiveness of the proposed PGUDA framework. The results demonstrate that our proposed PGUDA achieves leading performance in both cross-subject and cross-session classification tasks, achieving average accuracies of 58.08% and substantially outperforming existing DA approaches. Notably, PGUDA exhibits remarkable label efficiency: it attains classification accuracy comparable to fully supervised benchmarks while requiring only 5% of labeled data for teacher network training. This framework offers a robust and data-efficient solution that can significantly reduce the calibration burden in practical sEMG-based gesture recognition systems.

eess.SP↗

arXiv2026

DRIFT: Difficulty Routing Self-DIstillation with Rhythm-Gated Exploration and Success BuFfer Training

Enabling large language models to achieve stable self-improvement without external expert supervision remains a central challenge in complex reasoning tasks. Existing self-distillation and reinforcement learning methods lack explicit mechanisms for tracking problem-level learning progress and adapting optimization strategies accordingly. Consequently, training may over-optimize easy problems, receive weak supervision from hard problems, and fail to sufficiently explore borderline cases. To resolve these issues, we propose DRIFT, an online self-evolution policy optimization framework for large language models. DRIFT regulates the model's self-improvement process through the joint use of Difficulty Routing and Rhythm Gating. The former identifies the model's learning state at the problem level and dynamically allocates self-distillation and reinforcement learning signals, while the latter refines policy updates at the token level, concentrating exploration on critical reasoning positions. By further incorporating a success buffer and a two-stage curriculum learning strategy, DRIFT preserves high-quality historical experience while progressively guiding the model from reliable behavior acquisition toward stable policy evolution. Evaluated across five benchmarks and three model scales, DRIFT surpasses the peak performance of both GRPO and SDPO across all evaluated metrics. On the average score over the five benchmarks, DRIFT achieves 79.5$\%$, outperforming GRPO by 9.5$\%$ and SDPO by 7.5$\%$, establishing a new state-of-the-art result. Notably, on ToolUse, DRIFT reaches an accuracy of 79.2$\%$, improving over GRPO by 13.5$\%$ and SDPO by 10.7$\%$, setting a new state-of-the-art and substantially outperforming all concurrent methods.

cs.LG↗

arXiv2026

Machine learning models for estimating counterfactuals in a single-arm inflammatory bowel disease study

Single-arm trials accelerate study timelines by reducing the number of patients that must be recruited for a concurrent control group. However, these designs require an alternative comparator to estimate treatment effects. One approach is to construct a virtual control arm using a machine learning (ML) model trained on external control data to predict the counterfactual outcomes of the treatment arm. Our aim in this study was to leverage virtual controls by developing and evaluating ML-based counterfactual outcome models trained on IFX-treated patients to predict 1-year steroid-free clinical remission (SFCR ) and a composite of C-reactive protein remission plus steroid-free clinical remission (CRP-SFCR) for ADA-treated pediatric Crohn's disease patients, and to compare the resulting IFX-versus-ADA treatment effect estimates with those obtained using propensity score matching to external controls. Five ML models were used to train counterfactual models on the observed IFX cohort data. The resulting models were used to predict the counterfactual outcomes for the ADA arm patients. LGBM yields the best OR closest to the propensity score matched reference, and all 95% CI results align with the conclusion from the reference study that no statistical difference in the primary and secondary outcomes has been observed between the patients treated with ADA or IFX. Our study supports virtual controls as a viable and effective substitute for expensive, lengthy or unethical patient recruitment in an inflammatory bowel disease (IBD) trial. The developed gradient boosted prediction model can be used as a pretrained model to generate IFX counterfactual predictions in future studies, pending external validation and assessment of transportability.

cs.LG↗

arXiv2026

Reversable phase transitions in ferroic two-dimensional Nb2O2I4 through optically excited coherent phonons

We investigate optically induced phase transitions in the two-dimensional (2D) ferroelectric (FE) material Nb2O2I4 using real-time time-dependent density functional theory (rt-TDDFT). Our results demonstrate that tailored laser pulses can activate specific coherent phonon modes. Specifically, the anharmonic atomic distortions of the A1-1 and A1-2 modes at the {\Gamma}-point facilitate the reversal of in-plane polarization. By fine-tuning laser parameters, additional phonon modes at both the Y and {\Gamma} points are excited. The resulting nonequilibrium atomic dynamics enable the formation of previously unreported ferroic phases, including three antiferroelectric (AFE) phases and one ferrielectric (FiE) phase. Notably, these optically induced phases can be reverted to the initial FE state using appropriate techniques. This controllable reversibility among multiple ferroic phases positions 2D Nb2O2I4 as a highly promising candidate for next-generation electronic storage applications.

cond-mat.mtrl-sci↗

arXiv2026

Mapping generative AI use in the human brain: divergent neural, academic, and mental health profiles of functional versus socio emotional AI use

The widespread adoption of generative artificial intelligence conversational agents (AICAs) among university students constitutes a novel cognitive social environment whose impact on the maturing brain remains elusive. Combining surveys with high resolution structural MRI, we examined patterns of general, functional, and socio emotional AICA use, academic performance, mental health, and brain structural signatures in a comparatively large sample of 222 young individuals. Across computational anatomy, meta analytic network level, and behavioral decoding analyses, we observed use specific associations. Higher general and functional AICA use frequencies were linked to better academic outcomes (GPA), larger dorsolateral prefrontal and calcarine gray matter volume, and enhanced hippocampal network clustering and local efficiency. In contrast, more frequent socio emotional AICA use was associated with poorer mental health (depression, social anxiety) and lower volume of superior temporal and amygdalar regions central to social and affective processing. These findings indicate that the same class of AI tools exerts distinct effects depending on usage patterns and motivations, engaging prefrontal hippocampal systems that support cognition versus socio emotional systems that may track distress linked usage. These heterogeneities are crucial for designing environments that harness the educational benefits of AI while mitigating mental health risks.

q-bio.NC↗

arXiv2026

Non-Hermitian Mosaic Maryland model

We introduce the non-Hermitian mosaic Maryland model, where a discrete modulation period and a non-Hermitian phase are incorporated into the potential, rendering the originally exactly solvable system generally non-integrable. This model provides a unique platform to investigate how structural modulation governs localization in complex quasiperiodic potentials. Using Avila's global theory, we analytically derive the exact Lyapunov exponent and obtain explicit formulas for the complex mobility edges. Remarkably, for modulation periods kappa >= 2, the system intrinsically hosts kappa-1 robust extended bands that persist independently of the potential strength and non-Hermiticity. We further characterize the topological nature of these phases via the spectral winding number. Unlike the standard Maryland model, the mosaic modulation induces mobility edges, and the resulting phase transitions are continuous, reflecting the non-integrable nature of the system. Numerical calculations of the inverse participation ratio and fractal dimension confirm the analytical predictions for the asymptotic form of the mobility edges in the large non-Hermiticity limit. This work establishes structural design as a powerful degree of freedom for engineering wave transport and enhancing the robustness of extended states in non-Hermitian systems.

cond-mat.dis-nn↗

arXiv2026

Generative Recommendation for Large-Scale Advertising

Generative recommendation has recently attracted widespread attention in industry due to its potential for scaling and stronger model capacity. However, deploying real-time generative recommendation in large-scale advertising requires designs beyond large-language-model (LLM)-style training and serving recipes. We present a production-oriented generative recommender co-designed across architecture, learning, and serving, named GR4AD (Generative Recommendation for ADdvertising). As for tokenization, GR4AD proposes UA-SID (Unified Advertisement Semantic ID) to capture complicated business information. Furthermore, GR4AD introduces LazyAR, a lazy autoregressive decoder that relaxes layer-wise dependencies for short, multi-candidate generation, preserving effectiveness while reducing inference cost, which facilitates scaling under fixed serving budgets. To align optimization with business value, GR4AD employs VSL (Value-Aware Supervised Learning) and proposes RSPO (Ranking-Guided Softmax Preference Optimization), a ranking-aware, list-wise reinforcement learning algorithm that optimizes value-based rewards under list-level metrics for continual online updates. For online inference, we further propose dynamic beam serving, which adapts beam width across generation levels and online load to control compute. Large-scale online A/B tests show up to 4.2% ad revenue improvement over an existing DLRM-based stack, with consistent gains from both model scaling and inference-time scaling. GR4AD has been fully deployed in Kuaishou advertising system with over 400 million users and achieves high-throughput real-time serving.

cs.IR↗

arXiv2025

Causal-Tune: Mining Causal Factors from Vision Foundation Models for Domain Generalized Semantic Segmentation

Fine-tuning Vision Foundation Models (VFMs) with a small number of parameters has shown remarkable performance in Domain Generalized Semantic Segmentation (DGSS). Most existing works either train lightweight adapters or refine intermediate features to achieve better generalization on unseen domains. However, they both overlook the fact that long-term pre-trained VFMs often exhibit artifacts, which hinder the utilization of valuable representations and ultimately degrade DGSS performance. Inspired by causal mechanisms, we observe that these artifacts are associated with non-causal factors, which usually reside in the low- and high-frequency components of the VFM spectrum. In this paper, we explicitly examine the causal and non-causal factors of features within VFMs for DGSS, and propose a simple yet effective method to identify and disentangle them, enabling more robust domain generalization. Specifically, we propose Causal-Tune, a novel fine-tuning strategy designed to extract causal factors and suppress non-causal ones from the features of VFMs. First, we extract the frequency spectrum of features from each layer using the Discrete Cosine Transform (DCT). A Gaussian band-pass filter is then applied to separate the spectrum into causal and non-causal components. To further refine the causal components, we introduce a set of causal-aware learnable tokens that operate in the frequency domain, while the non-causal components are discarded. Finally, refined features are transformed back into the spatial domain via inverse DCT and passed to the next layer. Extensive experiments conducted on various cross-domain tasks demonstrate the effectiveness of Causal-Tune. In particular, our method achieves superior performance under adverse weather conditions, improving +4.8% mIoU over the baseline in snow conditions.

cs.CV↗

arXiv2025

Extreme Model Compression with Structured Sparsity at Low Precision

Deep neural networks (DNNs) are used in many applications, but their large size and high computational cost make them hard to run on devices with limited resources. Two widely used techniques to address this challenge are weight quantization, which lowers the precision of all weights, and structured sparsity, which removes unimportant weights while retaining the important ones at full precision. Although both are effective individually, they are typically studied in isolation due to their compounded negative impact on model accuracy when combined. In this work, we introduce SLOPE Structured Sparsity at Low Precision), a unified framework, to effectively combine structured sparsity and low-bit quantization in a principled way. We show that naively combining sparsity and quantization severely harms performance due to the compounded impact of both techniques. To address this, we propose a training-time regularization strategy that minimizes the discrepancy between full-precision weights and their sparse, quantized counterparts by promoting angular alignment rather than direct matching. On ResNet-18, SLOPE achieves $\sim20\times$ model size reduction while retaining $\sim$99% of the original accuracy. It consistently outperforms state-of-the-art quantization and structured sparsity methods across classification, detection, and segmentation tasks on models such as ResNet-18, ViT-Small, and Mask R-CNN.

cs.CV↗

arXiv2025

Pinpointing crucial steps: Attribution-based Credit Assignment for Verifiable Reinforcement Learning

While Reinforcement Learning with Verifiable Rewards (RLVR) enhances complex reasoning in LLMs, current methods struggle to balance exploration and exploitation. This leads to critical issues like inaccurate credit assignment for intermediate steps and premature entropy collapse, limiting model performance. To address this, we introduce Attribution-based Contribution to Policy Optimization (ACPO), a phased framework that incorporates a difficulty-aware curriculum. ACPO improves exploration by using trajectory semantic segmentation and an attribution-based representation to dynamically regulate policy entropy, thus mitigating its collapse. Concurrently, it enhances exploitation with a factorized reward system that precisely quantifies the hierarchical contribution of each reasoning step, ensuring accurate credit assignment. Extensive experiments on challenging benchmarks, including AIME, MATH, and AMC, demonstrate that ACPO significantly outperforms existing state-of-the-art approaches.

cs.LG↗

arXiv2025

UAV-VL-R1: Generalizing Vision-Language Models via Supervised Fine-Tuning and Multi-Stage GRPO for UAV Visual Reasoning

Recent advances in vision-language models (VLMs) have demonstrated strong generalization in natural image tasks. However, their performance often degrades on unmanned aerial vehicle (UAV)-based aerial imagery, which features high resolution, complex spatial semantics, and strict real-time constraints. These challenges limit the applicability of general-purpose VLMs to structured aerial reasoning tasks. To address these challenges, we propose UAV-VL-R1, a lightweight VLM explicitly designed for aerial visual reasoning. It is trained using a hybrid method that combines supervised fine-tuning (SFT) and multi-stage reinforcement learning (RL). We leverage the group relative policy optimization (GRPO) algorithm to promote structured and interpretable reasoning through rule-guided rewards and intra-group policy alignment. To support model training and evaluation, we introduce a high-resolution visual question answering dataset named HRVQA-VL, which consists of 50,019 annotated samples covering eight UAV-relevant reasoning tasks, including object counting, transportation recognition, and spatial scene inference. Experimental results show that UAV-VL-R1 achieves a 48.17% higher zero-shot accuracy than the Qwen2-VL-2B-Instruct baseline and even outperforms its 72B-scale variant, which is 36x larger, on multiple tasks. Ablation studies reveal that while SFT improves semantic alignment, it may reduce reasoning diversity in mathematical tasks. GRPO-based RL compensates for this limitation by enhancing logical flexibility and the robustness of inference. Additionally, UAV-VL-R1 requires only 3.9GB of memory under FP16 inference and can be quantized to 2.5GB with INT8, supporting real-time deployment on resource-constrained UAV platforms.

cs.CV↗

arXiv2025

Ordering According to Size of Disks in a Narrow Channel

A long and narrow channel confines disks of two sizes. The disks are randomly agitated in a widened channel under moderate pressure, then jammed according to a tunable protocol. We present exact results that characterize jammed macrostates (volume, entropy, jamming patterns). The analysis divides jammed disk sequences into overlapping tiles out of which statistically interacting quasiparticles are constructed. The fractions of small and large disks are controlled by a chemical potential adapted to configurational statistics of granular matter. The results show regimes for the energy parameters (determined by the jamming protocol) that either enhance or suppress the mixing of disk sizes. Size segregation or size alternation driven by steric forces alone are manifestations of a broken symmetry.

cond-mat.soft↗

arXiv2025

Jammed disks of two sizes in a channel: segregation driven by steric forces

Disks of two sizes are confined to a long and narrow channel. The axis and the plane of the channel are horizontal. The channel is closed off by pistons that freeze jammed microstates out of loose disk configurations, agitated randomly at calibrated intensity and subject to moderate pressure. Disk sizes and channel width are such that under jamming no disks remain loose and all disks touch one wall. The protocol permits disks to move past each other prior to jamming, which facilitates randomness in the sequence of large and small disks. We present exact results for the characterization of jammed macrostates including volume and entropy for given fractions of small and large disks as functions of energy parameters which depend on the jamming protocol. Our analysis divides the disk sequence of jammed microstates into overlapping tiles out of which we construct 17 species of statistically interacting quasiparticles. Jammed macrostates then depend on the fractions of small and large disks and on a dimensionless control parameter inferred from measures for expansion work against the pistons and intensity of random agitations. Two models are introduced for comparison of key technical aspects: one model emphasizes symmetry and the other mechanical stability. We distinguish regimes for the energy parameters that either enhance or suppress mixing of disk sizes in jammed macrostates. The latter case, if realizable, is a manifestation of grain segregation driven by steric forces alone, without directional bias.

cond-mat.soft↗

arXiv2025

MiniCPM4: Ultra-Efficient LLMs on End Devices

This paper introduces MiniCPM4, a highly efficient large language model (LLM) designed explicitly for end-side devices. We achieve this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems. Specifically, in terms of model architecture, we propose InfLLM v2, a trainable sparse attention mechanism that accelerates both prefilling and decoding phases for long-context processing. Regarding training data, we propose UltraClean, an efficient and accurate pre-training data filtering and generation strategy, and UltraChat v2, a comprehensive supervised fine-tuning dataset. These datasets enable satisfactory model performance to be achieved using just 8 trillion training tokens. Regarding training algorithms, we propose ModelTunnel v2 for efficient pre-training strategy search, and improve existing post-training methods by introducing chunk-wise rollout for load-balanced reinforcement learning and data-efficient tenary LLM, BitCPM. Regarding inference systems, we propose CPM.cu that integrates sparse attention, model quantization, and speculative sampling to achieve efficient prefilling and decoding. To meet diverse on-device requirements, MiniCPM4 is available in two versions, with 0.5B and 8B parameters, respectively. Furthermore, we construct a hybrid reasoning model, MiniCPM4.1, which can be used in both deep reasoning mode and non-reasoning mode. Evaluation results demonstrate that MiniCPM4 and MiniCPM4.1 outperform similar-sized open-source models across benchmarks, with the 8B variants showing significant speed improvements on long sequence understanding and generation.

cs.CL↗

arXiv2025

AgentDNS: A Root Domain Naming System for LLM Agents

The rapid evolution of Large Language Model (LLM) agents has highlighted critical challenges in cross-vendor service discovery, interoperability, and communication. Existing protocols like model context protocol and agent-to-agent protocol have made significant strides in standardizing interoperability between agents and tools, as well as communication among multi-agents. However, there remains a lack of standardized protocols and solutions for service discovery across different agent and tool vendors. In this paper, we propose AgentDNS, a root domain naming and service discovery system designed to enable LLM agents to autonomously discover, resolve, and securely invoke third-party agent and tool services across organizational and technological boundaries. Inspired by the principles of the traditional DNS, AgentDNS introduces a structured mechanism for service registration, semantic service discovery, secure invocation, and unified billing. We detail the architecture, core functionalities, and use cases of AgentDNS, demonstrating its potential to streamline multi-agent collaboration in real-world scenarios. The source code will be published on https://github.com/agentdns.

cs.AI↗