arXiv subjects

Zijie Liu

Publications and source records attributed to Zijie Liu.

At least 19 recordsLinked to original sources

arXiv2026

Teacher-guided multi-task learning for integrated plasma monitoring on EAST

This paper integrates tokamak plasma disruption prediction, Edge-Localized Mode (ELM) monitoring, Multifaceted Asymmetric Radiation From the Edge (MARFE) detection, and H/L-mode identification into a unified Multi-Task Learning (MTL) model. This integration aims to realize MTL's potential for improving task performance while reducing inference time. However, on datasets constructed from selected EAST discharges recorded between 2015 and 2020, this study found that the traditional MTL method underperformed the corresponding Single-Task Learning (STL) method on some tasks. This performance degradation is the well-known issue of negative transfer. To address this issue, this study proposes the teacher-guided MTL method, using task-specific STL ensembles (teachers) to guide the training of an MTL model (student). The ensemble teachers guide training through soft-target distillation losses for each task of the MTL student model, and increased weighting of each shot's peak-disruptivity window in the disruption prediction label-supervised loss. Test-set results showed that the teacher-guided MTL method outperformed the corresponding STL method in ROC-AUC for all four tasks, eliminating the negative transfer observed with the traditional MTL method. In addition, on the same single GPU or single physical CPU core, the MTL student model can substantially reduce the inference time required to generate all four task outputs compared with the four STL baseline models. This paper offers a promising route to plasma monitoring systems that cover a broader range of phenomena, improve per-task performance and reduce inference time with limited real-time computational resources, supporting future high-performance plasma operation.

physics.plasm-ph↗

arXiv2026

AgentOrch: How Close Can Global Orchestration Come to Step-by-Step Interaction?

Enterprise workflow agents typically rely on step-by-step interaction (Step-by-Step): the agent executes an action, observes its result, and invokes a large language model (LLM) again to decide the next step. Global orchestration (Global) instead generates a complete executable workflow in advance and executes it without further LLM calls, substantially reducing inference overhead, but it remains less accurate than Step-by-Step. We study how close Global can come to Step-by-Step and what narrows the remaining gap. We introduce AgentOrch, a benchmark that directly compares global orchestration with step-by-step interaction on 650 enterprise SOPs and 3,220 runtime variants under matched tools, environments, and evaluation. Step-by-Step remains more accurate for every evaluated model, but the gap has shrunk by about 79% from early models to the strongest recent model. We identify three routes to further narrowing this gap. Incremental workflow construction improves Global accuracy by an average of 9.98 percentage points (pp) without runtime execution feedback. Revising complete workflows with execution feedback improves all evaluated models, including 6.13 pp on held-out runtime variants. Targeted training on complete workflows improves accuracy by an average of 12.18 pp across all five trained models. Although a gap remains, global orchestration offers a promising path toward enterprise agents that deliver higher accuracy with fewer LLM calls.

cs.AI↗

arXiv2026

ReGE: Stronger Agents through Experience Synthesis

Repeated attempts at interactive tasks uncover facts, useful operations, and failures that can inform subsequent execution. We introduce ReGE, a framework for turning this recorded experience into guidance for a fresh attempt. Its main implementation, ReGE-Syn, uses a single coach call to jointly synthesize bounded execution records and available artifacts into self-contained guidance, connecting task-specific findings with proposed actions and relevant pitfalls. A student starts from the task's initial environment state and adapts this guidance to current observations. Source generation, synthesis, and student execution use the same base model, without parameter updates or a stronger teacher. Across four benchmarks, ReGE improves accuracy by an average of 9.9 percentage points for GPT and 8.8 percentage points for Claude compared with unguided execution. On matched BrowseComp-Plus subsets, ReGE-Syn reduces guidance-construction tokens by 98.2\% on GPT and 95.5\% on Claude relative to adapted AggAgent, with observed success. Combined construction and execution tokens decrease by 15.7\% and 35.4\%, respectively, counting input plus output and excluding source generation. BrowseComp-Plus ablations further associate access to process information with fewer student steps and show that joint synthesis produces substantially shorter guidance than Direct history or Independent summaries, with observed success. These results demonstrate how agents can use their own recorded experience to improve subsequent task completion, and suggest a low-cost way to consolidate the execution records of parallel agent systems such as agent swarms.

cs.AI↗

arXiv2026

DrugReason: Dynamic Multi-View Reasoning over Knowledge Graph and Language Evidence for Drug Repurposing

Drug repurposing aims to identify new therapeutic uses for existing compounds and, compared with de novo drug discovery, offers a faster and more cost-effective path to clinical translation. However, the space of candidate drug-disease pairs is enormous and their underlying relationships often depend on complex multi-hop biological mechanisms, making it difficult to reliably predict which pairs represent true therapeutic relationships. Existing approaches tackle this from two directions: knowledge graph-based methods organize curated biomedical evidence into structured relational networks for grounded multi-hop reasoning, while LLM-based methods leverage pretrained knowledge to generate flexible mechanistic rationales. Yet neither is sufficient alone - KGs are confined to observed graph structure while LLMs lack factual grounding and risk hallucination. To address this gap, we propose DrugReason, a multi-view reasoning framework that integrates grounded KG reasoning with LLM-generated mechanistic inference for drug repurposing. DrugReason adaptively routes diverse reasoning paths to specialized experts conditioned on the query context, while a cross-expert distillation objective enables knowledge sharing without sacrificing expert specialization. Experiments on PharmaDB, DDInter, and DrugBank show that DrugReason improves average performance over strong single-view reasoning baselines and achieves competitive or superior results compared with graph-based alternatives, while providing interpretable routing-based predictions.

cs.LG↗

arXiv2026

When Retain Constraints Conflict: Mitigating Forget-Retain Interference in Tabular Data

Machine unlearning aims to remove the influence of designated training data while preserving model utility, but its behavior on tabular data remains underexplored. This gap is important because tabular prediction is widely used in high-stakes domains and is increasingly adapted to language models through record serialization and schema-aware prompting. We identify a key challenge that distinguishes tabular unlearning from unlearning in free-form text or other modalities: schema-induced forget-retain overlap. In serialized tabular data, records share fixed column-name/value slots, similar attribute ranges, and common output spaces. Consequently, a forget row may have nearby retain rows that rely on the same high-signal attributes, causing retain preservation to oppose the update required for forgetting. Motivated by this failure mode, we propose Conflict-Aware Unlearning (CAU), a schema-aware approach that reduces forget-retain interference by relaxing preservation constraints on retained rows that most conflict with the forget set. Across sample-level and feature-level unlearning on clinical and non-medical tabular tasks, CAU more closely matches a retraining oracle while maintaining predictive utility and retain-region behavior. Our results show that reliable tabular LLM unlearning depends not only on the forgetting objective, but also on how retain constraints are constructed.

cs.LG↗

arXiv2026

Combating Catastrophic Forgetting in Continual Domain Adaptation via Knowledge Recasting

Adapting language models to new domains during post-training risks degrading previously acquired capabilities, a phenomenon commonly known as catastrophic forgetting. Existing mitigation methods largely treat forgetting as a consequence of excessive model drift. They therefore constrain updates by preserving old outputs, replaying old data, or regularizing changes to parameters and representations so that the updated model remains close to its previous state. We take a different view: Rather than pulling the updated model back toward its old state, we translate the new domain into the model's pre-existing knowledge space, making unfamiliar knowledge learnable through familiar procedures. Therefore, we propose Skill Schema Transport (SST), a post-training framework that transports new knowledge into the model's existing skill space. SST abstracts each example type into a skill schema, preserving its essential operations, dependencies, and verification steps while removing domain-specific surface forms. By recasting new examples as instances of reusable procedures, SST turns knowledge update from trajectory memorization into procedural reuse. In continual domain adaptation settings, SST consistently reduces forgetting while improving generalization to related tasks. Across two 3B instruction-tuned backbones, SST improves average final performance on streamed domains by 2.53 to 5.75 points over sequential post-training. It also lowers max-drop forgetting in all evaluated settings, with the largest reduction from 42.41 to 7.10 points on the Qwen-3B MATH stream, while substantially recovering the OOD degradation induced by sequential updates.

cs.LG↗

arXiv2026

Solving Large-Scale Parameter Inference in Controllable Nuclear Fusion via Coherent Ising Machines and Block Coordinate Descent

Bayesian inference enables self-consistent plasma parameter inversion in tokamak integrated data analysis by fusing multi-source diagnostics. However, under nonlinear Gaussian likelihoods, finding the maximum a posteriori (MAP) estimate typically relies on sampling methods whose high computational overhead precludes real-time plasma control. To address this bottleneck, we propose a Bayesian inversion framework using a Coherent Ising Machine (CIM)-a quantum computing architecture based on coupled degenerate optical parametric oscillators (DOPOs). With native all-to-all connectivity, the CIM directly maps the negative log-posterior potential onto a spin Hamiltonian and leverages quantum vacuum fluctuations for rapid state-space search during pitchfork bifurcation. Validated on plasma current profile inversion for the EAST tokamak, this method achieves a single-run solution time of approximately 1 ms. Although current qubit limits necessitate a block-wise iterative scheme that prevents full real-time execution today, rapidly scaling quantum hardware offers a clear pathway toward real-time tokamak control.

quant-ph↗

arXiv2026

Holo-World: Unified Camera, Object and Weather Control for Video World Model

Video world models are moving toward preserving an observed world under controllable camera and object motion while allowing its environmental state to change. Yet these controls remain isolated, and weather generation typically relies on a source video or reconstructed scene that already specifies future structure. We study a first-frame-anchored source-to-state setting, where the model starts from a single image and follows explicit camera and object controls and an optional weather instruction, then generates a video that either preserves the source world or transfers it to a target weather state. To address these challenges, we first build HoloStateData, a state video dataset that turns diverse videos into unified control samples for camera, object, and weather supervision. Second, we introduce Holo-World, a unified controllable video world model that jointly controls the scene from a single image. Its Unified Scene Adapter factorizes world preservation and weather transfer into distinct parameter subspaces, using rendered background, geometry buffers, and object controls to maintain controlled scene structure while modeling weather-dependent appearance and particle effects. Additionally, Scene-Weather Decomposed CFG guides scene and weather residuals separately, strengthening target weather effects without over-amplifying the full condition. Quantitative and qualitative experiments demonstrate that Holo-World maintains precise camera and object controls with consistent scene structure while transferring scenes into diverse target weather states, outperforming video-to-video weather editing baselines on weather-state generation. Our project page is available at https://xiangchenyin.github.io/Holo-World/.

cs.CV↗

arXiv2026

A 75-mL intrinsically stable atom-filtered laser enabling deployable quantum devices

Numerous quantum devices require lasers strictly locked to atomic transitions. Atom-filtered lasers (AFLs) are considered a leading candidate for quantum device laser sources due to their ability to self-align to atomic transitions. However, the sub-GHz sharp transmission spectra of conventional atomic filters impose constraints on both laser miniaturization and output stability. Herein, we demonstrate a micro Faraday anomalous dispersion optical filter (μFADOF) operating within the extreme hyperfine Paschen-Back regime, generating a 7.5 GHz flat-top transmission window. By integrating this filter, the miniaturization bottleneck of the AFLs is overcome, achieving a 30-fold volume reduction to a compact package volume of 75 mL. Simultaneously, investigations into the optical self-feedback characteristics of the μFADOF reveal a stable operating regime, where the optical self-feedback of the atomic filter acts as a stabilizing mechanism, enabling the laser to achieve a power instability of 9 x 10^-6 at 1 s and 3.2 x 10^-5 at 8400 s. The optical frequency standard based on this 75-mL AFL achieves a further improvement in long-term frequency stability to 3 x 10^-13 at 10000 s and maintains turn-key operation under environmental shocks, highlighting the critical role of this laser in the development of deployable quantum devices.

quant-ph↗

arXiv2026

State-Dependent Diffusion and Spectra of Strongly Driven Thermal Atoms

We propose a state-dependent diffusion model for strongly driven thermal-atom spectra. Starting from the trajectory-dependent internal-state evolution of individual atoms, we derive a closed spatial equation for the local density-matrix field using a velocity-moment expansion. Measurements of an $^{85}$Rb atomic-filter transmission spectrum agree well with the model up to a maximum Gaussian peak intensity of $1.27\times10^{3}$ W/cm$^2$, approaching six orders of magnitude above the $^{85}$Rb D2-line saturation intensity. Counterintuitively, the model reveals an anomalous optical-pumping pathway in which intense light transfers atoms from nominally dark states into bright states. Hyperfine Paschen--Back splitting selectively enhances this anomalous pathway while suppressing conventional optical pumping, allowing the filter to maintain approximately 97$\%$ transmission at the highest intensity studied. This work provides a framework for controlling strongly driven atomic ensembles and designing saturation-resistant atomic optical devices.

physics.atom-ph↗

arXiv2026

"very likely" Means "uncertain"? How LLMs Diverge from Humans in Linguistic Uncertainty Quantification

Humans express uncertainty verbally via markers (e.g., "possible," "likely"), yet most LLM uncertainty quantification (UQ) relies on costing likelihood- or consistency-based signals. From a cognitive perspective, accurate verbal uncertainty reflects metacognitive monitoring, representing knowledge boundaries ("knowing that you don't know") to support regulation and information seeking. In this paper, we investigate how LLMs diverge from humans in verbal uncertainty quantification and whether verbal markers can reliably quantify LLM uncertainty. We curate a corpus of human uncertainty markers from psychology and decision-science literature and benchmark LLMs against it. We observe that LLMs encode verbal uncertainty with numerical levels that differ substantially from those of humans. We then introduce METHODNAME, a novel optimization-based algorithm that learns an optimal uncertainty profile over uncertainty markers directly from LLM outputs. By fitting a marker-uncertainty mapping to best explain empirical correctness, METHODNAME discovers how much probability mass each verbal marker should convey, rather than estimating uncertainty via repeated sampling. METHODNAME enables a direct, marker-level comparison of confidence semantics between humans and LLMs, disentangling mismatch and revealing systematic confidence disparities in verbal expressions.

cs.LG↗

arXiv2026

Dual-Faraday-laser-pumped cesium beam clock with $7.7\times 10^{-13}/\sqrtτ$ frequency stability

Compact cesium beam clocks are major frequency references for deployable timing systems. However, further improvement of their short-term frequency stability is limited by the clock signal-to-noise ratio (SNR). Although two-laser optical pumping can increase the effective atomic utilization, the achievable clock SNR has long been limited by laser-induced frequency-to-amplitude noise conversion. Here, we demonstrate a compact dual-Faraday-laser-pumped (DFP) Cs beam clock enabled by a low-frequency-noise atom-referenced laser architecture. The intracavity Faraday anomalous dispersion optical filter provides inherent alignment to the Cs D$_2$ resonances, while modulation transfer spectroscopy offers suppressed frequency noise and drift. The resulting laser system supports robust turnkey operation with a Lorentzian linewidth of 2.12 kHz. The DFP Cs clock achieves a clock SNR of 46,365 in a 1-Hz bandwidth and a fractional Allan deviation of $7.7\times 10^{-13}/\sqrtτ$ , with Hadamard deviation reaching $7.7\times 10^{-15}$ at 10,000 s. This work pushes the fractional frequency stability of a compact Cs beam clock into the $10^{-13}/\sqrtτ$ regime, providing a pathway toward high-performance Cs frequency references for field-deployable precision timing, navigation, and synchronization.

physics.atom-ph↗

arXiv2026

Stochastic Meta-Unlearning: Bridging Language Backbone and Multimodal Unlearning

Machine unlearning for vision-language models (VLMs) remains underexplored. Unlike language models, VLMs combine a language backbone with visual components, which makes unlearning more complex. There is a surprising phenomenon when moving from single-modality unlearning to VLM unlearning: a target forgotten by the standalone language backbone can still be recovered when image information is given to the full VLM. This shows that text-only feedback is not enough for reliable VLM unlearning. Motivated by this observation, we propose Stochastic Meta-Unlearning (SMU), a bilevel framework that uses VLM-level feedback to learn an unlearning-ready initialization. In the inner loop, SMU applies a few unlearning steps to the language backbone using text data. In the outer loop, SMU recomposes the updated backbone with the frozen VLM and evaluates forgetting and utility at the VLM level. This design makes the unlearning update aware of the final multimodal behavior, while still keeping the update local to the language backbone. Experiments on two VLMs, two multimodal meme datasets, and three baselines show that SMU achieves the best overall forget-retain trade-off. Compared with the strongest baseline for each metric, SMU reduces average Forget accuracy by 10.52 points and improves average Retain and Test accuracy by 20.10 and 17.01 points, respectively. More importantly, SMU also transfers to new forgetting targets and to different meta-test unlearning methods. These results suggest that VLM-level feedback can make language-backbone unlearning more reliable and more transferable for VLMs.

cs.CL↗

arXiv2026

A Replicate-and-Quantize Strategy for Plug-and-Play Load Balancing of Sparse Mixture-of-Experts LLMs

Sparse Mixture-of-Experts (SMoE) architectures are increasingly used to scale large language models efficiently, delivering strong accuracy under fixed compute budgets. However, SMoE models often suffer from severe load imbalance across experts, where a small subset of experts receives most tokens while others are underutilized. Prior work has focused mainly on training-time solutions such as routing regularization or auxiliary losses, leaving inference-time behavior, which is critical for deployment, less explored. We present a systematic analysis of expert routing during inference and identify three findings: (i) load imbalance persists and worsens with larger batch sizes, (ii) selection frequency does not reliably reflect expert importance, and (iii) overall expert workload and importance can be estimated using a small calibration set. These insights motivate inference-time mechanisms that rebalance workloads without retraining or router modification. We propose Replicate-and-Quantize (R&Q), a training-free and near-lossless framework for dynamic workload rebalancing. In each layer, heavy-hitter experts are replicated to increase parallel capacity, while less critical experts and replicas are quantized to remain within the original memory budget. We also introduce a Load-Imbalance Score (LIS) to measure routing skew by comparing heavy-hitter load to an equal allocation baseline. Experiments across representative SMoE models and benchmarks show up to 1.4x reduction in imbalance with accuracy maintained within +/-0.6%, enabling more predictable and efficient inference.

cs.LG↗

arXiv2026

RAVEN: Agentic RAG for Automated Vulnerability Repair

Automated vulnerability repair has emerged as a promising direction to mitigate the growing number of software vulnerabilities. Recent advances in Large Language Models (LLMs) have further accelerated research in automated repair. However, existing frameworks remain largely restricted to memory-related vulnerabilities and locally repairable vulnerability settings, leaving generalization to unseen vulnerability types underexplored. Their evaluations are often limited to a single programming language, and largely rely on proprietary models. In this paper, we propose RAVEN, a scalable, efficient and autonomous framework that integrates an agentic retrieval-augmented generation (RAG) pipeline with controlled iterative repair in a unified framework. The framework utilizes open-source LLMs in a fully locally deployable setting with limited GPU requirements, while building a multi-faceted retrieval pipeline to retrieve historically relevant vulnerability fixes and guide the patch generation. In addition, RAVEN introduces a dedicated Curator Agent that retrieves cross-file dependencies from the target repository, to fix complex vulnerabilities that cannot be addressed using local vulnerable code alone. We evaluate RAVEN on 160 real-world CVE vulnerabilities across diverse vulnerability types, two programming languages, unseen CWE categories, and out-of-distribution settings. RAVEN achieves an overall repair success rate of 83.13%, outperforming all existing state-of-the-art repair frameworks, while also demonstrating strong generalization capabilities and maintaining the repair cost negligible.

cs.CR↗

arXiv2026

GEMQ: Global Expert-Level Mixed-Precision Quantization for MoE LLMs

Mixture-of-Experts Large Language Models (MoE-LLMs) achieve strong performance but incur substantial memory overhead due to massive expert parameters. Mixed-precision quantization mitigates this cost by allocating expert-wise bit-widths based on their importance, approaching the accuracy-memory Pareto frontier and enabling extreme low-bit quantization. However, existing methods rely on layer-wise importance estimation and overlook router shifts induced by quantization, resulting in suboptimal allocation and routing. In this work, we propose Global Expert-level Mixed-precision Quantization (GEMQ) to overcome these limitations via (1) a global linear-programming formulation that captures model-wide expert importance based on quantization error analysis, and (2) efficient router fine-tuning to adapt routing to quantized experts. These components are integrated into a progressive quantization framework that iteratively refines importance estimation and allocation. Experiments demonstrate that GEMQ significantly reduces memory and accelerates inference with minimal accuracy degradation. Source code is available at https://github.com/jndeng/GEMQ .

cs.LG↗

arXiv2026

Fairness for distribution network operations and planning

The incorporation of fairness into the distribution network (DN) planning and operation has become a key goal of recent studies. The cost of implementing fairness, denominated the price of fairness (PoF), covers the efficiency that is renounced for attaining social cohesion through fair outcomes. Locational disparity makes fairness schemes emerge to level the consumers playing field. However, fairness encompasses a range of notions. From egalitarian to merit-based criteria, various metrics are implemented as a tool for measuring equitable utility distribution. These have different mathematical complexities, from linear to non-linear programming cases, which affect their overall applicability. Hence, this study compiles the overarching fairness notions and metrics, reviewing how these affect stakeholders and the inherent mathematical optimisation in resource allocation problems. The aim is to support consistent and transparent planning and decision-making within DN operations.

cs.AI↗

arXiv2026

SWE-Next: Scalable Real-World Software Engineering Tasks for Agents

Executable software engineering data is valuable for training SWE agents, but scaling it remains difficult for two reasons: only a small fraction of real repository changes yield verifiable, high-signal task instances, and naively building repository-specific environments quickly becomes the dominant systems cost. We present SWE-Next, an execution-grounded framework for scalable SWE task and trajectory collection. On the data side, SWE-Next mines real merged pull requests, executes candidate base/merged commit pairs, and retains only those that produce strict test improvements without regressions, yielding self-verifying instances. It also applies strict submission gating so that collected trajectories remain evidence-driven rather than speculative. On the systems side, SWE-Next introduces reusable repo-quarter profiles, which reuse the same environment across nearby commits in time while keeping each task run separate and reproducible. Using only 30 hours and 639GB of environment storage, SWE-Next processes 3,971 seed repositories and 102,582 candidate commit pairs mined from real merged PRs to construct a dataset of 2,308 self-verifying instances. Experiments show that SWE-Next improves downstream pass@1 with fewer or comparable training trajectories, indicating that its gains come not from a stronger trajectory generator, but from higher-signal execution-grounded supervision and more efficient data collection.

cs.SE↗