arXiv subjects

Yanlong Zhao

Publications and source records attributed to Yanlong Zhao.

At least 19 recordsLinked to original sources

arXiv2026

SR-OPSD: Self-Referenced On-Policy Self-Distillation

On-policy self-distillation (OPSD) converts feedback into dense token-level supervision on trajectories generated by the policy to be optimized, providing a useful complement to reinforcement learning with sparse outcome rewards. However, the self-teacher policy used in OPSD is typically a stop-gradient or exponential-moving-average copy of the policy conditioned on additional context information, and thus co-evolves with both the student policy and its on-policy context distribution. Directly matching such a moving target with a fixed projection objective can lead to unstable optimization or excessive distributional concentration. This nature of OPSD motivates the proposed \emph{Self-Referenced On-Policy Self-Distillation (SR-OPSD)}. At fixed student-generated contexts, a token-level variational characterization identifies the effective distillation target as a geometric interpolation between the self-teacher policy and a reference policy. Meanwhile, we use the R\'enyi divergence family to generalize the projection geometry. This formulation separates \emph{where} the adaptive target is placed from \emph{how} the student is projected toward it: the interpolation coefficient controls underlying target, while the R\'enyi order controls the projection geometry and its sensitivity to token-level density ratios. Extensive experiments across scientific evaluation, mathematical reasoning, and coding generation tasks with multiple large language models show that SR-OPSD achieves the state-of-the-art or competitive performance across various settings.

cs.LG↗

arXiv2026

Saliency-Aware Regularized Quantization Calibration for Large Language Models

Post-training quantization (PTQ) is an effective approach for deploying large language models (LLMs) under memory and latency constraints. Most existing PTQ methods determine quantization parameters by minimizing a layer-wise reconstruction error on a predetermined calibration dataset, typically optimized via either scale search or Gram-based methods. However, from the perspective of generalization risk, existing PTQ calibration objectives based solely on empirical reconstruction error over limited or unrepresentative calibration data may move the quantized weights away from the original floating-point weights, potentially degrading downstream performance. To address this issue, we propose \emph{Regularized Quantization Calibration} (RQC), a unified framework that augments standard PTQ objectives with a regularizer that explicitly controls weight deviation from the original weights. We further generalize this framework to incorporate a saliency-aware regularizer, resulting in \emph{Saliency-Aware Regularized Quantization Calibration} (SARQC). The proposed regularization encourages quantized weights to remain close to the original weights during calibration, leading to improved generalization at inference time. SARQC integrates seamlessly into existing PTQ pipelines and enhances both scale-search-based and Gram-based methods under a unified formulation. Extensive experiments on dense and Mixture-of-Experts LLMs demonstrate consistent improvements in perplexity and zero-shot accuracy, without introducing additional inference overhead.

cs.AI↗

arXiv2026

Asymptotically Efficient Recursive Identification Under One-Bit Communications Achieving Original CRLB

This paper develops an asymptotically efficient recursive identification algorithm for autoregressive systems with exogenous inputs under one-bit communications. In particular, the proposed method asymptotically achieves the Cramer-Rao lower bound (CRLB) based on the original data before quantization (original CRLB), whereas existing approaches typically attain only the CRLB corresponding to the quantized observations. The primary reason is that the existing methods quantize only the current system output, resulting in non-negligible information loss under one-bit quantization. To overcome this challenge, we present a novel quantization method that integrates both current and historical system outputs and inputs to provide richer parameter information in one-bit data, allowing the information loss caused by quantization to become a minor term relative to the original CRLB. Based on this technique, a corresponding remote estimation algorithm is further proposed. To address the convergence analysis challenge posed by the non-independence of the one-bit data, we establish a new framework that analyzes the tail probability of integrated data formed by combining current and historical system outputs and inputs before quantization, thereby eliminating the need for the traditional independence assumption on the quantized data. It is proven that the remote estimate achieves asymptotic normality, and the error covariance matrix converges to the original CRLB, confirming its asymptotic efficiency. Compared to existing identification algorithms under one-bit data, this method reduces the asymptotic mean squared error by at least 36%. Several numerical examples are simulated to show the effectiveness of the proposed algorithm.

math.OC↗

arXiv2026

Recursive Binary Identification with Differential Privacy and Data Tampering Attacks

In this paper, we consider the parameter estimation in a bandwidth-constrained sensor network communicating through an insecure medium. The sensor performs a local quantization, and transmits a 1-bit message to an estimation center through a wireless medium where the transmission of information is vulnerable to attackers. Both eavesdroppers and data tampering attackers are considered in our setting. A differential privacy method is used to protect the sensitive information against eavesdroppers. Then, a recursive projection algorithm is proposed such that the estimation center achieves the almost sure convergence and mean-square convergence when quantized measurements, differential privacy, and data tampering attacks are considered in a uniform framework. A privacy analysis including the convergence rate with privacy or without privacy is given. Further, we extend the problem to multi-agent systems. For this case, a distributed recursive projection algorithm is proposed with guaranteed almost sure and mean square convergence. A simulation example is provided to illustrate the effectiveness of the proposed algorithms.

eess.SY↗

arXiv2025

Distributed Fusion Estimation with Protecting Exogenous Inputs

In the context of distributed fusion estimation, directly transmitting local estimates to the fusion center may cause a privacy leakage concerning exogenous inputs. Thus, it is crucial to protect exogenous inputs against full eavesdropping while achieving distributed fusion estimation. To address this issue, a noise injection strategy is provided by injecting mutually independent noises into the local estimates transmitted to the fusion center. To determine the covariance matrices of the injected noises, a constrained minimization problem is constructed by minimizing the sum of mean square errors of the local estimates while ensuring ({\epsilon}, {\delta})-differential privacy. Suffering from the non-convexity of the minimization problem, an approach of relaxation is proposed, which efficiently solves the minimization problem without sacrificing differential privacy level. Then, a differentially private distributed fusion estimation algorithm based on the covariance intersection approach is developed. Further, by introducing a feedback mechanism, the fusion estimation accuracy is enhanced on the premise of the same ({\epsilon}, {\delta})-differential privacy. Finally, an illustrative example is provided to demonstrate the effectiveness of the proposed algorithms, and the trade-off between differential privacy level and fusion estimation accuracy.

eess.SY↗

arXiv2025

DualHNIE: Dual-Channel Hypergraph Learning for Node Importance Estimation in Heterogeneous Knowledge Graphs

Estimating node importance in heterogeneous knowledge graphs is a fundamental problem underlying recommendation, search, and knowledge decision systems. However, most existing methods rely on pairwise message passing mechanisms that fail to capture higher-order interactions induced by meta-relational structures. Furthermore, structural topology and semantic attributes are typically entangled within a unified embedding space, which obscures their distinct inductive biases and limits the discriminative capacity of learned importance representations. To address these limitations, we propose DualHNIE, a principled dual-channel hypergraph learning framework for node importance estimation. DualHNIE first constructs a higher-order knowledge graph by forming typed hyperedges from meta-path sequences, enabling explicit modeling of higher-order relational patterns. It then introduces two complementary encoders: a structure-aware hypergraph attention network that performs locally normalized aggregation over meta-path--induced hyperedges to capture localized structural dependencies, and a sparse--chunked hypergraph transformer that captures global semantic interactions while maintaining scalable computation. We further design a contrastive alignment mechanism with auxiliary supervision, ensuring cross-view consistency while preserving modality-specific representation. Extensive experiments on multiple benchmark datasets demonstrate that DualHNIE outperforms state-of-the-art methods, validating the effectiveness of explicit high-order modeling and disentangled dual-channel representation learning for heterogeneous knowledge graphs. Code and datasets are available\footnote[1]{https://github.com/jiawenchen10/DualHNIE}.

cs.AI↗

arXiv2025

Quantized Distributed Estimation with Event-triggered Communication and Packet Loss

This paper focuses on the problem of quantized distributed estimation with event-triggered communication and packet loss, aiming to reduce the number of transmitted bits. The main challenge lies in the inability to differentiate between an untriggered event and a packet loss occurrence. This paper proposes an event-triggered distributed estimation algorithm with quantized communication and quantized measurement, in which it introduces a one-bit information reconstruction method to deal with packet loss. The almost sure convergence and convergence rate of the proposed algorithm are established. Besides, it is demonstrated that the global average communication bit-rate decreases to zero over time. Moreover, the trade-off between communication rate and convergence rate is revealed, providing guidance for designing the communication rate required to achieve the algorithm's convergence rate. A numerical example is supplied to validate the findings.

eess.SY↗

arXiv2025

Recursive Binary Identification under Data Tampering and Non-Persistent Excitation with Application to Emission Control

This paper studies the problem of online parameter estimation for cyber-physical systems with binary outputs that may be subject to adversarial data tampering. Existing methods are primarily offline and unsuitable for real-time learning. To address this issue, we first develop a first-order gradient-based algorithm that updates parameter estimates recursively using incoming data. Considering that persistent excitation (PE) conditions are difficult to satisfy in feedback control scenarios, a second-order quasi-Newton algorithm is proposed to achieve faster convergence without requiring the PE condition. For both algorithms, corresponding versions are developed to handle known and unknown tampering strategies, and their parameter estimates are proven to converge almost surely over time. In particular, the second-order algorithm ensures convergence under a signal condition that matches the minimal excitation required by classical least-squares estimation in stochastic regression models. The second-order algorithm is also extended to an adaptive control framework, providing an explicit upper bound on the tracking error for binary-output FIR systems under unknown tampering. Three numerical simulations verify the theoretical results and show that the proposed methods are robust against data tampering. Finally, the approach is validated via a vehicle emission control problem, where it effectively improves the detection accuracy of excess-emission events.

eess.SY↗

arXiv2025

Scale-Disentangled spatiotemporal Modeling for Long-term Traffic Emission Forecasting

Long-term traffic emission forecasting is crucial for the comprehensive management of urban air pollution. Traditional forecasting methods typically construct spatiotemporal graph models by mining spatiotemporal dependencies to predict emissions. However, due to the multi-scale entanglement of traffic emissions across time and space, these spatiotemporal graph modeling method tend to suffer from cascading error amplification during long-term inference. To address this issue, we propose a Scale-Disentangled Spatio-Temporal Modeling (SDSTM) framework for long-term traffic emission forecasting. It leverages the predictability differences across multiple scales to decompose and fuse features at different scales, while constraining them to remain independent yet complementary. Specifically, the model first introduces a dual-stream feature decomposition strategy based on the Koopman lifting operator. It lifts the scale-coupled spatiotemporal dynamical system into an infinite-dimensional linear space via Koopman operator, and delineates the predictability boundary using gated wavelet decomposition. Then a novel fusion mechanism is constructed, incorporating a dual-stream independence constraint based on cross-term loss to dynamically refine the dual-stream prediction results, suppress mutual interference, and enhance the accuracy of long-term traffic emission prediction. Extensive experiments conducted on a road-level traffic emission dataset within Xi'an's Second Ring Road demonstrate that the proposed model achieves state-of-the-art performance.

cs.LG↗

arXiv2025

Set-Valued Transformer Network for High-Emission Mobile Source Identification

Identifying high-emission vehicles is a crucial step in regulating urban pollution levels and formulating traffic emission reduction strategies. However, in practical monitoring data, the proportion of high-emission state data is significantly lower compared to normal emission states. This characteristic long-tailed distribution severely impedes the extraction of discriminative features for emission state identification during data mining. Furthermore, the highly nonlinear nature of vehicle emission states and the lack of relevant prior knowledge also pose significant challenges to the construction of identification models.To address the aforementioned issues, we propose a Set-Valued Transformer Network (SVTN) to achieve comprehensive learning of discriminative features from high-emission samples, thereby enhancing detection accuracy. Specifically, this model first employs the transformer to measure the temporal similarity of micro-trip condition variations, thus constructing a mapping rule that projects the original high-dimensional emission data into a low-dimensional feature space. Next, a set-valued identification algorithm is used to probabilistically model the relationship between the generated feature vectors and their labels, providing an accurate metric criterion for the classification algorithm. To validate the effectiveness of our proposed approach, we conducted extensive experiments on the diesel vehicle monitoring data of Hefei city in 2020. The results demonstrate that our method achieves a 9.5\% reduction in the missed detection rate for high-emission vehicles compared to the transformer-based baseline, highlighting its superior capability in accurately identifying high-emission mobile pollution sources.

cs.LG↗

arXiv2025

Asymptotic Efficiency Analysis of the Recursive Least-Squares Algorithm for ARX Systems Without Projection

This paper investigates the optimality analysis of the recursive least-squares (RLS) algorithm for autoregressive systems with exogenous inputs (ARX systems). A key challenge in analyzing is managing the potential unboundedness of the parameter estimates, which may diverge to infinity. Previous approaches addressed this issue by assuming that both the true parameter and the RLS estimates remain confined within a known compact set, thereby ensuring uniform boundedness throughout the analysis. In contrast, we propose a new analytical framework that eliminates the need for such a boundness assumption. Specifically, we establish a quantitative relationship between the bounded moment conditions of quasi-stationary input/output signals and the convergence rate of the tail probability of the RLS estimation error. Based on this technique, we prove that when system inputs/outputs have bounded twentieth-order moments, the RLS algorithm achieves asymptotic normality and the covariance matrix of the RLS algorithm converges to the Cram\'er-Rao lower bound (CRLB), confirming its asymptotic efficiency. These results demonstrate that the RLS algorithm is an asymptotically optimal identification algorithm for ARX systems, even without the projection operators to ensure that parameter estimates reside within a prior known compact set.

math.OC↗

arXiv2025

Consensus Tracking Control of Multi-agent Systems with A Time-varying Reference State under Binary-valued Communication

This paper investigates the problem of consensus tracking control of discrete time multi-agent systems under binary-valued communication. Different from most existing studies on consensus tracking, the transmitted information between agents is the binary-valued. Parameter identification with binary-valued observations is applied to the estimation of neighbors'states and the tracking control is designed based on the estimation. Two Lyapunov functions are constructed to deal with the strong coupling of estimation and control. Compared with consensus problems under binary-valued communication, a reference state is required for consensus tracking control. Two scenarios of the time-varying reference state are studied respectively. (1) The reference state is asymptotically convergent. An online algorithm that performs estimation and control simultaneously is proposed, in which the estimation step size and the control gain are decreasing with time. By this algorithm, the multi-agent system is proved to achieve consensus tracking with convergence rate O(1/k^{\epsilon} ) under certain conditions. (2) The reference state is bounded, which is less conservative than that in the first case. In this case, the estimation step size and control gain are designed to be constant. By this algorithm, all the followers can reach to a neighborhood of the leader with an exponential rate. Finally, simulations are given to demonstrate theoretical results.

cs.MA↗

arXiv2025

One-bit consensus of controllable linear multi-agent systems with communication noises

This paper addresses the one-bit consensus of controllable linear multi-agent systems (MASs) with communication noises. A consensus algorithm consisting of a communication protocol and a consensus controller is designed. The communication protocol introduces a linear compression encoding function to achieve a one-bit data rate, thereby saving communication costs. The consensus controller with a stabilization term and a consensus term is proposed to ensure the consensus of a potentially unstable but controllable MAS. Specifically, in the consensus term, we adopt an estimation method to overcome the information loss caused by one-bit communications and a decay step to attenuate the effect of communication noise. Two combined Lyapunov functions are constructed to overcome the difficulty arising from the coupling of the control and estimation. By establishing similar iterative structures of these two functions, this paper shows that the MAS can achieve consensus in the mean square sense at the rate of the reciprocal of the iteration number under the case with a connected fixed topology. Moreover, the theoretical results are generalized to the case with jointly connected Markovian switching topologies by establishing a certain equivalence relationship between the Markovian switching topologies and a fixed topology. Two simulation examples are given to validate the algorithm.

eess.SY↗

arXiv2025

A Unified Recursive Identification Algorithm with Quantized Observations Based on Weighted Least-Squares Type Criteria

This paper investigates system identification problems with Gaussian inputs and quantized observations under fixed thresholds. By reinterpreting the nonlinear effects induced by quantization as the product of the unknown parameter and an unknown nonlinear coefficient, this work establishes a novel weighted least-squares criterion that enables linear estimation of unknown parameters under quantized observations. Subsequently, a two-step recursive identification algorithm is constructed by estimating two unknown terms, which is capable of handling both Gaussian noisy and noise-free linear systems. Convergence analysis of this identification algorithm is conducted, demonstrating convergence in both almost sure and $L^{p}$ senses under mild conditions, with respective rates of $O(\sqrt{ \log \log k/k})$ and $O(1/k^{p/2})$, where $k$ denotes the time step. In particular, this algorithm offers an asymptotically efficient estimation of the variance of Gaussian variables using quantized observations. Furthermore, extensions to output-error systems are discussed, enhancing the applicability and relevance of the proposed methods. Two numerical examples are provided to validate these theoretical advancements.

math.OC↗

arXiv2024

Distributed Estimation with Quantized Measurements and Communication over Markovian Switching Topologies

This paper addresses distributed parameter estimation in stochastic dynamic systems with quantized measurements, constrained by quantized communication and Markovian switching directed topologies. To enable accurate recovery of the original signal from quantized communication signal, a persistent excitation-compliant linear compression encoding method is introduced. Leveraging this encoding, this paper proposes an estimation-fusion type quantized distributed identification algorithm under a stochastic approximation framework. The algorithm operates in two phases: first, it estimates neighboring estimates using quantized communication information, then it creates a fusion estimate by combining these estimates through a consensus-based distributed stochastic approximation approach. To tackle the difficulty caused by the coupling between these two estimates, two combined Lyapunov functions are constructed to analyze the convergence performance. Specifically, the mean-square convergence of the estimates is established under a conditional expectation-type cooperative excitation condition and the union topology containing a spanning tree. Besides, the convergence rate is derived to match the step size's order under suitable step-size coefficients. Furthermore, the impact of communication uncertainties including stochastic communication noise and Markov-switching rate is analyzed on the convergence rate. A numerical example illustrates the theoretical findings and highlights the joint effect of sensors under quantized communication.

eess.SY↗

arXiv2024

Recursive Projection-Free Identification with Binary-Valued Observations

This paper is concerned with parameter identification problem for finite impulse response (FIR) systems with binary-valued observations under low computational complexity. Most of the existing algorithms under binary-valued observations rely on projection operators, which leads to a high computational complexity of much higher than O(n^2). In response, this paper introduces a recursive projection-free identification algorithm that incorporates a specialized cut-off coefficient to fully utilize prior information, thereby eliminating the need for projection operators. The algorithm is proved to be mean square and almost surely convergent. Furthermore, to better leverage prior information, an adaptive accelerated coefficient is introduced, resulting in a mean square convergence rate of O(1/k) , which matches the convergence rate with accurate observations. Inspired by the structure of the Cramer-Rao lower bound, the algorithm can be extended to an information-matrix projection-free algorithm by designing adaptive weight coefficients. This extension is proved to be asymptotically efficient for first-order FIR systems, with simulations indicating similar results for high order FIR systems. Finally, numerical examples are provided to demonstrate the main results.

eess.SY↗

arXiv2024

State Estimation with Protecting Exogenous Inputs via Cram\'er-Rao Lower Bound Approach

This paper addresses the real-time state estimation problem for dynamic systems while protecting exogenous inputs against adversaries, who may be honest-but-curious third parties or external eavesdroppers. The Cram\'er-Rao lower bound (CRLB) is employed to constrain the mean square error (MSE) of the adversary's estimate for the exogenous inputs above a specified threshold. By minimizing the MSE of the state estimate while ensuring a certain privacy level measured by CRLB, the problem is formulated as a constrained optimization. To solve the optimization problem, an explicit expression for CRLB is first provided. As the computational complexity of the CRLB increases with the time step, a low-complexity approach is proposed to make the complexity independent of time. Then, a relaxation approach is proposed to efficiently solve the optimization problem. Finally, a privacy-preserving state estimation algorithm with low complexity is developed, which also ensures $(\epsilon, \delta)$-differential privacy. Two illustrative examples, including a practical scenario for protecting building occupancy, demonstrate the effectiveness of the proposed algorithm.

eess.SY↗

arXiv2024

Signal-Comparison-Based Distributed Estimation Under Decaying Average Data Rate Communications

The paper investigates the distributed estimation problem under low bit rate communications. Based on the signal-comparison (SC) consensus protocol under binary-valued communications, a new consensus+innovations type distributed estimation algorithm is proposed. Firstly, the high-dimensional estimates are compressed into binary-valued messages by using a periodic compressive strategy, dithered noises and a sign function. Next, based on the dithered noises and expanding triggering thresholds, a new stochastic event-triggered mechanism is proposed to reduce the communication frequency. Then, a modified SC consensus protocol is applied to fuse the neighborhood information. Finally, a stochastic approximation estimation algorithm is used to process innovations. The proposed SC-based algorithm has the advantages of high effectiveness and low communication cost. For the effectiveness, the estimates of the SC-based algorithm converge to the true value in the almost sure and mean square sense. A polynomial almost sure convergence rate is also obtained. For the communication cost, the local and global average bit rates for communications decay to zero at a polynomial rate. The trade-off between the convergence rate and the communication cost is established through event-triggered coefficients. A better convergence rate can be achieved by decreasing event-triggered coefficients, while lower communication cost can be achieved by increasing event-triggered coefficients. A simulation example is given to demonstrate the theoretical results.

eess.SY↗