arXiv subjects

Hanwen Li

Publications and source records attributed to Hanwen Li.

9 recordsLinked to original sources

arXiv2026

STRATA: Self-Maintaining Structured Experience Graphs for Self-Evolving Language Agents

Large language model (LLM) agents can self-evolve at test time by accumulating experience: retrieving relevant memories before acting and writing new lessons back afterwards. Yet most are single-level: trajectories, distilled workflows and reasoning items all land in one undifferentiated pool indexed by embedding similarity, with no typed abstraction hierarchy. They never forget or deduplicate, and are prone to negative transfer, where a superficially similar but directionally opposite experience is recalled and its concrete steps are copied verbatim, actively harming the agent. We introduce STRATA, a structured, outcome-grounded, self-maintaining memory that organizes experience into a three-layer graph of atomic skills and code snippets, task workflows and generalized meta-frameworks, connected by typed edges. STRATA maintains itself through five mechanisms sharing one similarity geometry: a multi-dimensional similarity that scores retrieval by operation and target; a typed reflector crystallizing procedures from successes and pitfalls from failures; community detection that induces meta-frameworks; pressure-triggered garbage collection with articulation-point protection; and contrast edges that flag confusable, direction-opposite experiences. Across four interactive benchmarks (AppWorld, WebArena, ALFWorld, ScienceWorld), two frozen backbones and seven baselines, STRATA attains the best average score while simultaneously using the lowest latency and fewest environment steps of any memory-augmented method, and 44% fewer lines of injected memory than the most context-hungry baseline. We isolate two failure modes of prior memories: single-level memory degrades sharply as the task set grows, and single-level concrete memories suffer catastrophic negative transfer, which contrast edges repair, lifting ScienceWorld from 78.7 to 96.7 within the same system.

cs.AI↗

arXiv2026

UMAS: System-Level Uncertainty Quantification for Multi-Agent LLM Systems

Multi-Agent Systems (MAS) are a promising paradigm for improving the reasoning capabilities of large language models, yet they remain prone to hallucinations and erroneous outputs. Uncertainty Quantification (UQ) is crucial for assessing system reliability, but existing methods are largely designed for single-agent settings; how to exploit MAS structure for UQ while keeping inference cost-neutral remains underexplored. We propose UMAS, a parameter-free, cost-neutral, system-level UQ framework that estimates the credibility of inter-agent influence from each generation's intrinsic confidence. UMAS uses these credibility signals to modulate how uncertainty is updated at each node, and then aggregates node-level uncertainty into a system-level estimate. Extensive evaluations on Debate and DyLAN across diverse benchmarks show that UMAS consistently outperforms state-of-the-art baselines by an average of 10.54 AUROC points. Beyond uncertainty estimation, UMAS enables hallucination detection and uncertainty-aware answer selection, improving MAS accuracy by up to 12.6 points on specific tasks while enhancing reliability. Code is available at https://github.com/Dby52-04/UMAS.

cs.MA↗

arXiv2026

From Uncertainty to Action: Learning to Steer LLM Agents

Steering an LLM agent means deciding whether to correct it, at which step, and with which mechanism. Uncertainty is often used to decide when to correct an agent, but whether it can guide these decisions remains unclear. We steer agent trajectories separately at every non-terminal step with each of four mechanisms and run each continuation to completion. The resulting stepwise outcome table (SOT) holds about 82,000 counterfactual continuations of 1,864 trajectories from three benchmarks and two agents. It shows that uncertainty can identify failing trajectories, but that no single signal reliably locates the step at which steering helps. We therefore propose VoS (Value of Steering), a trajectory-level monitor, offline or online, that learns from SOT the value of steering at each step and decides where to steer by it. A harm-budgeted trigger decides whether to steer, limiting the fraction of successful trajectories that VoS disturbs. VoS improves on unmodified execution in all 12 settings of benchmark, agent, and offline or online use, by 7.8 points on average, and outperforms the strongest of five existing uncertainty-triggered methods in 11, by 2.9 points on average. Ablations show that training on measured outcomes and a tight harm budget are both essential.

cs.AI↗

arXiv2025

Pockels effect induced strong Kerr nonlinearity in a lithium niobate waveguide

The utilization of Kerr nonlinearity in lithium niobate has been extensively investigated over the years. Nevertheless, the practical implementation of Kerr nonlinearity in waveguides has been constrained by the material's inherently low third-order nonlinear coefficients. Here, we present a significant advancement by demonstrating Pockels effect-induced strong Kerr nonlinearity in a periodically poled thin-film lithium niobate waveguide. Both effective four-wave mixing (FWM) and cascaded effective FWM processes are experimentally observed. The induced FWM process achieves a remarkable maximum output power of -8.5 dBm, spanning a wavelength spectrum of over 116.8 nm. Analysis reveals that the induced effective Kerr nonlinearity exhibits a substantial effective nonlinear refractive index as $2.9\times 10^{-15} m^{2}W^{-1}$, corresponding to an effective nonlinear refractive index enhancement factor of $1.6\times 10^{4}$ relative to the intrinsic value. Moreover, a wavelength-converting experiment demonstrates a flat optic-to-optic response over a broadband radiofrequency spectrum, confirming that signal integrity is well preserved after on-chip effective FWM conversion. Therefore, the demonstrated efficient and broadband Pockels effect induced effective Kerr nonlinearity paves the way for novel applications in diverse fields, including spectroscopy, parametric amplification, quantum correlation studies, and wavelength conversion technologies.

physics.optics↗

arXiv2024

PZT Optical Memristors

Optical memristors represent a monumental leap in the fusion of photonics and electronics, heralding a new era of applications from neuromorphic computing to artificial intelligence. However, current technologies are hindered by complex fabrication, limited endurance, high optical loss or low modulation depth. For the first time, we reveal optical non-volatility in thin-film Lead Zirconate Titanate (PZT) by electrically manipulating the ferroelectric domains to control the refractive index, providing a brand-new routine for optical memristors. The developed PZT optical memristors offer unprecedented advantages more than exceptional performance metrics like low loss of <2 dB/cm, high precision exceeding 6-bits, large modulation depth with an index change as large as 4.6x10-3. Additionally, these devices offer impressive stability, maintaining minimal wavelength variation for over three weeks and enduring more than 10,000 cycles, and require a mere 0.8 pJ of energy for non-volatile operation. The wafer-scale sol-gel fabrication process also ensures compatible with standardized mass fabrication processes and high scalability for photonic integration. Specially, these devices also demonstrate unique functional duality: setting above a threshold voltage enables non-volatile behaviors, below this threshold allows volatile high-speed optical modulation. This marks the first-ever optical memristor capable of performing high-speed (48 Gbps) and energy-efficient (450 fJ/bit) signal processing and non-volatile retention on a single platform, and is also the inaugural demonstration of scalable functional systems. The PZT optical memristors developed here facilitate the realization of novel paradigms for high-speed and energy-efficient optical interconnects, programmable PICs, quantum computing, neural networks, in-memory computing and brain-like architecture.

physics.optics↗

arXiv2022

A modified gravity model based on network efficiency for vital nodes identification in complex networks

Vital nodes identification is an essential problem in network science. Various methods have been proposed to solve this problem. In particular, based on the gravity model, a series of improved gravity models are proposed to find vital nodes better in complex networks. However, they still have the room to be improved. In this paper, a novel and improved gravity model, which is named network efficiency gravity centrality model (NEG), integrates gravity model and network efficiency is proposed. Compared to other methods based on different gravity models, the proposed method considers the effect of the nodes on structure robustness of the network better. To solidate the superiority of the proposed method, experiments on varieties of real-world networks are carried out.

cs.SI↗

arXiv2022

DVS: Deep Visibility Series and its Application in Construction Cost Index Forecasting

Time series forecasting is a hot spot in recent years. Visibility Graph (VG) algorithm is used for time series forecasting in previous research, but the forecasting effect is not as good as deep learning prediction methods such as methods based on Artificial Neural Network (ANN), Convolutional Neural Network (CNN) and Long Short-Term Memory Network (LSTM). The visibility graph generated from specific time series contains abundant network information, but the previous forecasting method did not effectively use the network information to forecast, resulting in relatively large prediction errors. To optimize the forecasting method based on VG, this article proposes the Deep Visibility Series (DVS) module through the bionic design of VG and the expansion of the past research. By applying the bionic design of biological vision to VG, DVS has obtained superior forecasting accuracy. At the same time, this paper applies the DVS forecasting method to the construction cost index forecast, which has practical significance.

cs.LG↗

arXiv2022

Evaluating importance of nodes in complex networks with local volume information dimension

How to evaluate the importance of nodes is essential in research of complex network. There are many methods proposed for solving this problem, but they still have room to be improved. In this paper, a new approach called local volume information dimension is proposed. In this method, the sum of degree of nodes within different distances of central node is calculated. The information within the certain distance is described by the information entropy. Compared to other methods, the proposed method considers the information of the nodes from different distances more comprehensively. For the purpose of showing the effectiveness of the proposed method, experiments on real-world networks are implemented. Promising results indicate the effectiveness of the proposed method.

cs.SI↗

arXiv2021

Uncertainty Measurement of Basic Probability Assignment Integrity Based on Approximate Entropy in Evidence Theory

Evidence theory is that the extension of probability can better deal with unknowns and inaccurate information. Uncertainty measurement plays a vital role in both evidence theory and probability theory. Approximate Entropy (ApEn) is proposed by Pincus to describe the irregularities of complex systems. The more irregular the time series, the greater the approximate entropy. The ApEn of the network represents the ability of a network to generate new nodes, or the possibility of undiscovered nodes. Through the association of network characteristics and basic probability assignment (BPA) , a measure of the uncertainty of BPA regarding completeness can be obtained. The main contribution of paper is to define the integrity of the basic probability assignment then the approximate entropy of the BPA is proposed to measure the uncertainty of the integrity of the BPA. The proposed method is based on the logical network structure to calculate the uncertainty of BPA in evidence theory. The uncertainty based on the proposed method represents the uncertainty of integrity of BPA and contributes to the identification of the credibility of BPA.

cs.AI↗