Searcharxiv⌕ Search

arXiv subjects

Nan Yang

Publications and source records attributed to Nan Yang.

At least 37 records · Page 2Linked to original sources
arXiv2026

On Training Large Language Models for Long-Horizon Tasks: An Empirical Study of Horizon Length

Large language models (LLMs) have shown promise as interactive agents that solve tasks through extended sequences of environment interactions. While prior work has primarily focused on system-level optimizations or algorithmic improvements, the role of task horizon length in shaping training dynamics remains poorly understood. In this work, we present a systematic empirical study that examines horizon length through controlled task constructions. Specifically, we construct controlled tasks in which agents face identical decision rules and reasoning structures, but differ only in the length of action sequences required for successful completion. Our results reveal that increasing horizon length alone constitutes a training bottleneck, inducing severe training instability driven by exploration difficulties and credit assignment challenges. We demonstrate that horizon reduction is a key principle to address this limitation, stabilizing training and achieving better performance in long-horizon tasks. Moreover, we find that horizon reduction is related to stronger generalization across horizon lengths: models trained under reduced horizons generalize more effectively to longer-horizon variants at inference time, a phenomenon we refer to as horizon generalization.

cs.AI↗
arXiv2026

Only Say What You Know: Calibration-Aware Generation for Long-Form Factuality

Large Reasoning Models achieve strong performance on complex tasks but remain prone to hallucinations, particularly in long-form generation where errors compound across reasoning steps. Existing approaches to improving factuality, including abstention and factuality-driven optimization, follow a \emph{coupled exploration-commitment} paradigm, in which intermediate reasoning is unconditionally propagated to the final output, limiting fine-grained control over information selection and integration. In this paper, we propose an \textbf{Exploration-Commitment Decoupling} paradigm that disentangles knowledge exploration from final commitment, enabling models to explore with awareness while answering cautiously. We instantiate the paradigm with \textbf{Calibration-Aware Generation (CAG)}, a framework that equips models with end-to-end, calibration-aware generation capabilities, by augmenting intermediate reasoning with calibrated reliability estimates and prioritizing reliable content in final outputs. Across five long-form factuality benchmarks and multiple model families, CAG improves factuality by up to 13%, while reducing decoding time by up to 37%. Overall, our work highlights decoupling as a principled approach for more reliable long-form generation, offering directions for trustworthy and self-aware generative systems.

cs.CL↗
arXiv2026

Artificial-Noise Aided Design for Movable-Antenna Enabled Physical-Layer Service Integration

This paper pioneers a novel scheme for artificial-noise (AN)-aided movable-antenna (MA)-enabled physical-layer service integration (PLSI) to harmonize the simultaneous delivery of multicast and confidential messages. By jointly exploiting the spatial reconfiguration capability of MAs and the interference shaping capability of AN, we aim to enhance secrecy performance while guaranteeing multicast reliability. The joint design of MA positions and transmit variables results in a highly coupled and non-convex optimization problem. To address this, we first provide key insights into the role of spatial degrees of freedom in AN design. We then characterize the AN direction under a structured transmission design and derive a closed-form expression for the AN-to-confidential power allocation ratio, which significantly simplifies the overall design. To solve the resulting problem, we further develop a low-complexity block coordinate ascent (BCA)-based scheme that alternates between transmit design and MA position optimization. Numerical results demonstrate that the proposed scheme achieves significant secrecy performance gains with low computational complexity and fast convergence, highlighting its effectiveness for MA-enabled PLSI systems.

cs.IT↗
arXiv2026

Gravitational waves of extreme-mass-ratio inspirals in a rotating black hole with Dehnen dark matter halo

Extreme Mass Ratio Inspirals (EMRIs) are among the key targe sources for the space-based gravitational wave (GW) detectors. The waveforms of the EMRIs are highly sensitive to the types of the central supermassive black hole (SBH) and can serve as a novel sensitive tool to probe the background spacetime. In this work, we compute GWs radiated from EMRIs in the backgrounds of Kerr black hole and rotating black hole with Dehnen-type dark matter halo (DMBH). Following the Teukolsky prescription, we obtain the perturbed equations for curvature tensor from the Newman-Penrose (NP) equations, and for the DMBH we obtain the radial and angular equations through separation of variables. To solve the equations with numerical method we apply the Sasaki-Nakamura (SN) transformation to convert the Teykolsky-type equation into the SN equation. We study the radiation reaction of GWs by computing the energy flux and angular momentum flux at infinity and at the horizon. The orbital evolution is then derived from the total fluxes. We extract the two polarizations of GWs by solving the equation numerically. By comparing the waveforms of Kerr and DMBH, it is found that the DM halo induces noticeable changes in both the amplitude and phase of GWs. We compute the strain of GW detector with the response function and evaluate the mismatch between the waveforms of Kerr and DMBH. The results show that the mismatch increases with the mass parameter of DM halo and the spin of the SBH.

gr-qc↗
arXiv2026

$H_2/H_{\infty}$ Control for Stochastic Differential Systems with Partial Observation

This paper investigates the $H_{2}/H_{\infty}$ control problem for linear stochastic differential systems under partial observation. Unlike existing studies that assume full state accessibility, we consider the scenario where the controller has access only to an observation process. The objective is to design a controller that balances the $H_2$ performance criterion with the $H_\infty$ robustness requirement under worst-case disturbances, formulated as a nonzero-sum differential game. Using the Kalman filtering method, we derive the corresponding optimal filtering equation. Furthermore, a Stochastic Bounded Real Lemma under the partial observation framework is established, providing necessary and sufficient conditions for the $H_\infty$ robustness constraint. We also show the connection between the existence of a Nash equilibrium and the solvability of the cross-coupled Riccati equations, and illustrate the effectiveness of the proposed approach through a numerical example involving an unmanned aerial vehicle (UAV).

math.OC↗
arXiv2026

CroSatFL: Energy-Efficient Federated Learning with Cross-Aggregation for Satellite Edge Computing

Low Earth Orbit (LEO) mega-constellations extend the cloud-to-edge continuum into space, enabling satellite edge computing. However, Federated Learning (FL) in this environment is fundamentally energy-constrained due to dynamic inter-satellite connectivity, heterogeneous onboard computing hardware, and strict power budgets. We propose CroSatFL, a sustainable on-orbit hierarchical FL framework that reduces end-to-end energy across computation and communication while maintaining strong training performance under realistic LEO dynamics. CroSatFL keeps the ground station (GS) off the iterative loop by performing all local training and intermediate aggregations on orbit, requiring only two GS communication phases: one for initialization and one for final model collection. This sharply reduces repeated use of bandwidth-limited and energy-expensive GS links and shifts iterative exchanges to laser inter-satellite links (LISLs). CroSatFL integrates three energy-aware mechanisms: StarMask forms LISL-feasible clusters that align data volume with heterogeneous CPU/GPU capability, Skip-One mitigates transient stragglers by skipping at most one slow client per cluster to lower round energy and latency while preserving long-term fairness, and random-k cross-aggregation enables lightweight topology-aware cross-cluster mixing without extending round duration. Using an end-to-end energy model with a realistic Walker-Delta constellation, we show that CroSatFL reduces GS communication count by over two orders of magnitude and GS transmission energy by about 6x relative to GS-centric and on-orbit baselines, while achieving competitive accuracy and faster convergence.

cs.DC↗
arXiv2026

Two Pathways to Truthfulness: On the Intrinsic Encoding of LLM Hallucinations

Despite their impressive capabilities, large language models (LLMs) frequently generate hallucinations. Previous work shows that their internal states encode rich signals of truthfulness, yet the origins and mechanisms of these signals remain unclear. In this paper, we demonstrate that truthfulness cues arise from two distinct information pathways: (1) a Question-Anchored pathway that depends on question-answer information flow, and (2) an Answer-Anchored pathway that derives self-contained evidence from the generated answer itself. First, we validate and disentangle these pathways through attention knockout and token patching. Afterwards, we uncover notable and intriguing properties of these two mechanisms. Further experiments reveal that (1) the two mechanisms are closely associated with LLM knowledge boundaries; and (2) internal representations are aware of their distinctions. Finally, building on these insightful findings, two applications are proposed to enhance hallucination detection performance. Overall, our work provides new insight into how LLMs internally encode truthfulness, offering directions for more reliable and self-aware generative systems.

cs.CL↗
arXiv2026

Distributed Optimization-Learning with Graph Transformers for Terahertz Cell-Free Integrated Sensing and Communication Systems

In this paper, we propose a distributed optimization-learning framework for terahertz (THz) cell-free integrated sensing and communication (CF-ISAC) systems, termed Distributed Optimization-Learning with Graph Transformers (DOLG). We first formulate a highly non-convex joint scheduling and signal design problem for THz CF-ISAC systems, jointly optimizing access point (AP)-user equipment (UE) association and beamforming under signal to interference plus noise ratio based communication and Cramér-Rao bound based sensing constraints, together with line-of-sight-driven visibility rules and per-AP power constraints. We also develop an optimization based benchmark utilizing a tractable relaxed reformulation. Building upon this optimization structure, we redesign a graph transformer network (GTN) as an optimization-aware representation module that encodes cross-field wavefront geometry, blockage visibility, and sensing relevance in a permutation-equivariant manner. The proposed DOLG framework amortizes the iterative optimization procedure into a scalable GTN-conditioned distributed multi-agent reinforcement learning policy through centralized training and decentralized execution, while preserving per-AP power constraints via structure-preserving projections. Simulation results demonstrate that the proposed DOLG framework achieves stable convergence and effectively balances the communication-sensing tradeoff. From the system-level perspective, it outperforms multicell and non-joint design baselines. Furthermore, it surpasses conventional optimization based and heuristic approaches in terms of both ISAC performance and computational scalability.

eess.SP↗
arXiv2026

DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance

Large Language Models (LLMs) are increasingly deployed as Internet/Web services (LLM-as-a-Service) with strict latency Service-Level Objectives (SLOs) under tight GPU memory budgets. Mixture-of-Experts (MoE) models improve quality and throughput via sparse expert activation, but serving them efficiently is challenging because expert weights dominate memory footprint and incur costly host--device transfers when offloaded. Moreover, MoE serving exhibits a phase disparity: the prefill phase tends to activate experts densely across many tokens, while the decode phase activates only a few experts per step. A uniform expert loading/caching policy across phases leads to either peak-memory blowup (prefill) or tail-latency inflation (decode). We present DuoServe-MoE, a QoS-oriented MoE serving system that decouples prefill and decode and applies phase-specialized expert scheduling. For prefill, DuoServe-MoE uses a two-stream CUDA pipeline to overlap expert prefetching with non-MoE computation, reducing expert residency time and peak GPU memory. For decode, it employs a lightweight layer-level predictor trained offline from activation traces to prefetch only likely experts without model changes. Experiments on representative MoE LLMs show that DuoServe-MoE improves TTFT by up to $5.34\times$ and end-to-end latency by up to $7.55\times$ over representative baselines, while maintaining low runtime GPU memory usage under resource-constrained deployment.

cs.DC↗
arXiv2026

AgentServe: Algorithm-System Co-Design for Efficient Agentic AI Serving on a Consumer-Grade GPU

Large language models (LLMs) are increasingly deployed as AI agents that operate in short reasoning-action loops, interleaving model computation with external calls. Unlike traditional chat applications, these agentic workloads require inference serving systems to balance low latency, stable token emission, and throughput under multiple request arrivals from different AI agents. Recent deployments highlight a shift toward running small language models (SLMs) locally on consumer-grade GPUs, driven by privacy, compliance, and cost constraints. When heterogeneous requests overlap on a single GPU, long prefills and short decodes contend for resources, creating head-of-line blocking that destabilizes interactive performance. By analyzing agent workloads, we observe that their execution naturally separates into cold prefills, which process long system prompts, resume prefills, which append tool outputs to cached contexts, and short decodes, which are latency-critical. This mix intensifies contention compared to conventional chatbot serving. We present AgentServe, a single-GPU serving system that ensures stable multi-agent execution under such conditions by isolating prefills from decodes, applying dynamic budgeting to resume prefills, and allocating GPU resources through pre-established CUDA Green Context slots with adaptive control. Evaluation results show that AgentServe significantly improves latency stability while sustaining competitive throughput, achieving up to 2.8x TTFT improvement and 2.7x TPOT improvement over state-of-the-art baselines across different settings.

cs.DC↗
arXiv2026

Learning to Draft: Adaptive Speculative Decoding with Reinforcement Learning

Speculative decoding accelerates large language model (LLM) inference by using a small draft model to generate candidate tokens for a larger target model to verify. The efficacy of this technique hinges on the trade-off between the time spent on drafting candidates and verifying them. However, current state-of-the-art methods rely on a static time allocation, while recent dynamic approaches optimize for proxy metrics like acceptance length, often neglecting the true time cost and treating the drafting and verification phases in isolation. To address these limitations, we introduce Learning to Draft (LTD), a novel method that directly optimizes for throughput of each draft-and-verify cycle. We formulate the problem as a reinforcement learning environment and train two co-adaptive policies to dynamically coordinate the draft and verification phases. This encourages the policies to adapt to each other and explicitly maximize decoding efficiency. We conducted extensive evaluations on five diverse LLMs and four distinct tasks. Our results show that LTD achieves speedup ratios ranging from 2.24x to 4.32x, outperforming the state-of-the-art method Eagle3 up to 36.4%.

cs.CL↗
arXiv2026

Impact of Pointing Errors and Correlated Wall Blockages on Practical Grid-based Indoor Terahertz Communication Systems

Terahertz (THz) communications has emerged as a promising technology for future wireless systems due to its potential to support extremely high data rates. However, severe path loss, blockage effects, and sensitivity to beam misalignment pose major challenges to reliable indoor THz communications. In this paper, we investigate the coverage probability of downlink transmission in a three-dimensional (3D) indoor THz communication system under structured access point (AP) deployments, with a focus on square and hexagonal grid topologies. A tractable analytical framework is developed to jointly account for human blockages, correlated wall blockages across APs, beam training, and residual pointing error. Numerical results demonstrate that wall blockage correlation significantly reduces the association and coverage probabilities, and its impact cannot be neglected in system performance analysis. Compared with square grid AP deployments, hexagonal grids consistently achieve higher coverage by mitigating correlated wall blockage effects and reducing the distances between user equipments (UEs) and their associated APs. Furthermore, coverage performance is shown to strongly depend on the UE location, with noticeable degradation as the UE moves away from its nearest AP. Residual pointing error is found to introduce substantial coverage loss, especially for longer links. In addition, beam training analysis reveals a non-monotonic relationship between antenna array size and training overhead, highlighting an inherent tradeoff among antenna configuration, beamwidth selection, and beam training efficiency. These findings provide useful insights into the design and deployment of practical indoor THz communication systems.

cs.IT↗
arXiv2026

Symbol-Aware Precoder Design for Physical-Layer Anonymous Communications

Physical-layer characteristics, such as channel state information (CSI) and transmitter noise induced by hardware impairments, are often uniquely associated with a transmitter. This paper investigates transmitter anonymity at the physical layer from a signal design perspective. We consider an anonymous communication problem where the receiver should reliably decode the signal from the transmitter but should not make use of the signal to infer the transmitter's identity.Transmitter anonymity is quantified using a Kullback-Leibler divergence (KLD)-based metric, which enables the formulation of explicit anonymity constraints in the precoder design.We then propose an anonymous symbol-level precoding strategy that preserves reliable communication under spatial multiplexing while preventing transmitter identification. The proposed framework employs a partitioned equal-gain combining (P-EGC) scheme that leverages receiver diversity without requiring transmitter-specific CSI. Simulation results demonstrate anonymity-reliability tradeoffs across different signal-to-noise ratios (SNRs) and numbers of data streams. Moreover, the results reveal opposite trends of anonymity with respect to transmitter-dependent noise variations in the low-SNR and high-SNR regimes.

eess.SP↗
arXiv2026

Impact of Pointing Error on Coverage Performance of 3D Indoor Terahertz Communication Systems

In this paper, we develop a tractable analytical framework for a three-dimensional (3D) indoor terahertz (THz) communication system to theoretically assess the impact of the pointing error on its coverage performance. Specifically, we model the locations of access points (APs) using a Poisson point process, human blockages as random cylinder processes, and wall blockages through a Boolean straight line process. A pointing error refers to beamforming gain and direction mismatch between the transmitter and receiver. We characterize it based on the inaccuracy of location estimate. We then analyze the impact of this pointing error on the received signal power and derive a tractable expression for the coverage probability, incorporating the multi-cluster fluctuating two-ray distribution to accurately model small-scale fading in THz communications. Aided by simulation results, we corroborate our analysis and demonstrate that the pointing error has a pronounced impact on the coverage probability. Specifically, we find that merely increasing the antenna array size is insufficient to improve the coverage probability and mitigate the detrimental impact of the pointing error, highlighting the necessity of advanced estimation techniques in THz communication systems.

eess.SP↗
arXiv2026

Optimal Look-back Horizon for Time Series Forecasting in Federated Learning

Selecting an appropriate look-back horizon remains a fundamental challenge in time series forecasting (TSF), particularly in the federated learning scenarios where data is decentralized, heterogeneous, and often non-independent. While recent work has explored horizon selection by preserving forecasting-relevant information in an intrinsic space, these approaches are primarily restricted to centralized and independently distributed settings. This paper presents a principled framework for adaptive horizon selection in federated time series forecasting through an intrinsic space formulation. We introduce a synthetic data generator (SDG) that captures essential temporal structures in client data, including autoregressive dependencies, seasonality, and trend, while incorporating client-specific heterogeneity. Building on this model, we define a transformation that maps time series windows into an intrinsic representation space with well-defined geometric and statistical properties. We then derive a decomposition of the forecasting loss into a Bayesian term, which reflects irreducible uncertainty, and an approximation term, which accounts for finite-sample effects and limited model capacity. Our analysis shows that while increasing the look-back horizon improves the identifiability of deterministic patterns, it also increases approximation error due to higher model complexity and reduced sample efficiency. We prove that the total forecasting loss is minimized at the smallest horizon where the irreducible loss starts to saturate, while the approximation loss continues to rise. This work provides a rigorous theoretical foundation for adaptive horizon selection for time series forecasting in federated learning.

cs.LG↗
arXiv2025

Fun-ASR Technical Report

In recent years, automatic speech recognition (ASR) has witnessed transformative advancements driven by three complementary paradigms: data scaling, model size scaling, and deep integration with large language models (LLMs). However, LLMs are prone to hallucination, which can significantly degrade user experience in real-world ASR applications. In this paper, we present Fun-ASR, a large-scale, LLM-based ASR system that synergistically combines massive data, large model capacity, LLM integration, and reinforcement learning to achieve state-of-the-art performance across diverse and complex speech recognition scenarios. Moreover, Fun-ASR is specifically optimized for practical deployment, with enhancements in streaming capability, noise robustness, code-switching, hotword customization, and satisfying other real-world application requirements. Experimental results show that while most LLM-based ASR systems achieve strong performance on open-source benchmarks, they often underperform on real industry evaluation sets. Thanks to production-oriented optimizations, Fun-ASR achieves state-of-the-art performance on real application datasets, demonstrating its effectiveness and robustness in practical settings. The code and models are accessible at https://github.com/FunAudioLLM/Fun-ASR .

cs.CL↗
arXiv2025

Synthetic Forgetting without Access: A Few-shot Zero-glance Framework for Machine Unlearning

Machine unlearning aims to eliminate the influence of specific data from trained models to ensure privacy compliance. However, most existing methods assume full access to the original training dataset, which is often impractical. We address a more realistic yet challenging setting: few-shot zero-glance, where only a small subset of the retained data is available and the forget set is entirely inaccessible. We introduce GFOES, a novel framework comprising a Generative Feedback Network (GFN) and a two-phase fine-tuning procedure. GFN synthesises Optimal Erasure Samples (OES), which induce high loss on target classes, enabling the model to forget class-specific knowledge without access to the original forget data, while preserving performance on retained classes. The two-phase fine-tuning procedure enables aggressive forgetting in the first phase, followed by utility restoration in the second. Experiments on three image classification datasets demonstrate that GFOES achieves effective forgetting at both logit and representation levels, while maintaining strong performance using only 5% of the original data. Our framework offers a practical and scalable solution for privacy-preserving machine learning under data-constrained conditions.

cs.LG↗
arXiv2025

Scaling Law Analysis in Federated Learning: How to Select the Optimal Model Size?

The recent success of large language models (LLMs) has sparked a growing interest in training large-scale models. As the model size continues to scale, concerns are growing about the depletion of high-quality, well-curated training data. This has led practitioners to explore training approaches like Federated Learning (FL), which can leverage the abundant data on edge devices while maintaining privacy. However, the decentralization of training datasets in FL introduces challenges to scaling large models, a topic that remains under-explored. This paper fills this gap and provides qualitative insights on generalizing the previous model scaling experience to federated learning scenarios. Specifically, we derive a PAC-Bayes (Probably Approximately Correct Bayesian) upper bound for the generalization error of models trained with stochastic algorithms in federated settings and quantify the impact of distributed training data on the optimal model size by finding the analytic solution of model size that minimizes this bound. Our theoretical results demonstrate that the optimal model size has a negative power law relationship with the number of clients if the total training compute is unchanged. Besides, we also find that switching to FL with the same training compute will inevitably reduce the upper bound of generalization performance that the model can achieve through training, and that estimating the optimal model size in federated scenarios should depend on the average training compute across clients. Furthermore, we also empirically validate the correctness of our results with extensive training runs on different models, network settings, and datasets.

cs.LG↗