Search arXiv⌕ Search

arXiv subjects

Mingyuan Li

Publications and source records attributed to Mingyuan Li.

At least 19 recordsLinked to original sources
arXiv2026

Diversity Combining for Multi-Path LLM Reasoning

Multi-path reasoning methods such as self-consistency (SC) sample $K$ reasoning paths and choose the most frequent answer. However, their gains quickly plateau as $K$ increases, and locating this saturation point currently requires fitting a scaling model to accuracies measured at several values of $K$ or stopping online per query. We formalize multi-path LLM reasoning as a diversity combining problem from wireless communications: each path is a noisy channel observation, and the pairwise correlation of path correctness caps the design-effect effective sample size of the vote at a finite ceiling. Generalized least squares (GLS) analysis shows that, for paths with equal gains, the optimal linear combiner of latent embeddings is uniform exactly when every row of the path covariance has the same sum, as under exchangeability, supporting majority vote as the natural default in standard SC while leaving room for weighting or pruning under heterogeneous prompt-template branches. Across 5 models and 12 benchmarks, prompt-template diversity reduces path correlation in $55$ of $60$ model-benchmark cells, with a mean relative reduction of $16\%$ and reductions in every task domain. We derive an Adaptive-K rule that uses a four-path pilot to select $K^*$, retaining $96$--$103\%$ of MV@$K{=}32$ accuracy across Math, QA, and NLU.

cs.AI↗
arXiv2026

Frozen Memory Is Not Enough: Rethinking External Memory as Extraction

Methods for improving knowledge use in large language models typically fall into two regimes. Non-parametric retrieval offers flexible access to external knowledge, but adds retrieval latency, context overhead, and only shallow integration with the backbone. Parametric adaptation is efficient at inference time, but entangles knowledge with model weights and can be hard to update, audit, or transfer. Engram-style hashed memory occupies a middle regime: it stores learned information in an external, addressable table, yet consumes that table through a small learned reader. This raises a basic question: when such a memory is moved across backbones, what matters more, the frozen memory itself or the target-side reader? We study this question through cross-model frozen-memory extraction, in which a memory trained on a source model is frozen and attached to a different target model, with only a lightweight reader trained. Ablations show that learned memory content and correct addressing both matter, while the amount of usable signal depends on how well the reader is aligned to the target model. In downstream question answering tasks, a dual-layer, four-branch reader nearly closes the gap between same-model and cross-model reuse, achieving an average score of 38.8 under our controlled evaluation protocol. Moreover, when the provider reader is directly compatible with the target interface, the frozen artifact can provide substantial utility without target-side training, while optional reader adaptation yields further improvement. These results suggest that Engram can serve as a reusable external knowledge artifact, provided that the target has access to a compatible reader interface; target-side adaptation can further improve alignment when direct reader reuse is insufficient.

cs.CL↗
arXiv2026

Who Verifies the Verifiers? Accountable Semantic Consensus for AI-Mediated Web Information Verification

Large language models (LLMs) increasingly decide whether Web evidence supports online claims, often by voting over several assessors whose errors are presumed independent and benign. On the open Web, retrieved pages can steer the assessors that read them, and assessors from one model family fail together. We prove that even for independent assessors whose every output sequence is possible on both valid and invalid claims, any protocol that decides only from their outputs and always reaches an agreed verdict never rejects invalid claims or rejects valid ones with positive probability. We present Astral, which pins all assessors to one committed snapshot of the Web evidence. With up to f assessors faulty, Astral accepts a verdict only when 2f+1 assessors attest to it above a confidence threshold and no signed challenge stands. On five public benchmarks with two of seven assessors adversarial, Astral lowers pooled attack success to 21.5-53.1%, against 54.7-77.8% for majority voting. The cost is coverage, as without attack 16.0-61.7% of claims receive a binary verdict and the rest are disputed for audit. Our code runs at https://anonymous.4open.science/r/Astral-55F1.

cs.CR↗
arXiv2026

HARMONY: Hierarchical Anchor Retrieval on Manifold for Oblivious-source Machinery Acoustic Anomaly Detection

In standard acoustic anomaly detection (ASD), models are usually trained separately for each machine or domain using explicit metadata. However, in realistic deployments, machine identifiers are incomplete, unreliable, or simply unavailable. Thus, metadata-free machinery ASD asks a single model to monitor mixed machine populations without machine IDs. In this paper, we argue that the main failure mode is not a weak detector family, but representation entanglement. The heterogeneous normal modes overlap in frozen pre-trained feature spaces, making anomaly scoring and approximate retrieval unreliable. We study this problem directly and propose HARMONY (Hierarchical Anchor Retrieval on Manifold for Oblivious-source acoustic aNomalY detection), an anchor-guided geometric repartitioning framework that reorganizes mixed-source features into compact Voronoi regions and reuses this structure for hierarchical retrieval. On DCASE 2020, HARMONY improves mean AUC from 75.40% to 90.00% (14.60 percentage points) over the strongest unified baseline and from 83.90% to 90.00% over a strong foundation-feature baseline. On a single CPU, its hierarchical retrieval achieves a 7.8x speedup over exhaustive search with only a 0.20 percentage-point AUC drop. These results suggest that incorporating explicit geometric partitioning can improve both detection accuracy and retrieval efficiency for ASD. Code is available at https://github.com/Zzyh0208/HARMONY.

cs.SD↗
arXiv2026

CD-OPD: Contextual Digest for On-Policy Distillation

Teacher takeover in on-policy distillation (OPD) can produce incorrect responses even when the student and teacher independently answer the same problem correctly, a failure we quantify using Collaboration Failure Rate (CFR). To address it, we propose Contextual Digest for On-Policy Distillation (CD-OPD). CD-OPD structures the earlier context of teacher guidance through a teacher-generated dependency graph, curating the required problem conditions and preceding reasoning as node summaries alongside the unchanged guidance in a Contextual Digest before student continuation. We further develop online meta-reweighting, which compares Digest segment gradients with a task gradient from independent student responses to improve student-only performance, favoring updates predicted to reduce task loss. Across eight benchmarks, CD-OPD improves average accuracy by 1.32--7.05 percentage points over representative existing OPD methods. It also reduces CFR by 27.4\%--56.4\% relative to existing takeover-based OPD methods.

cs.LG↗
arXiv2026

Outcome-Anchored Credit Reallocation for Agentic Knowledge Graph Question Answering

Agentic knowledge graph question answering (KGQA) requires large language models (LLMs) to solve complex questions through multi-turn interactions with a knowledge graph (KG). Outcome-supervised reinforcement learning derives verifiable outcome rewards from final answers but uniformly assigns each rollout's outcome-level advantage to all policy-generated tokens, creating a trajectory-to-operation mismatch since KG operations within a rollout can have varying effects on the search process and final answer. To address this, we propose Outcome-Anchored Credit Reallocation (OACR), which employs a training-only KG-state outcome model to estimate outcome-reward changes between adjacent KG search states and converts them into operation credit, from which token-wise weights are derived. These multiplicative weights reallocate token-level update magnitudes while preserving the sign of the outcome-level advantage at every generated token. Across three benchmarks, OACR achieves stronger overall performance than existing KGQA methods while using fewer inference-time LLM calls and tokens.

cs.AI↗
arXiv2026

OpCredit: Operation-Aware Policy Learning for Agentic Knowledge Graph Question Answering

Agentic knowledge graph question answering(KGQA) relies on sequences of executable operations, but trajectory-level rewards obscure their individual contributions. Successful trajectories may contain ineffective operations, while failed trajectories may contain useful ones. To assess individual operations, we propose OpCredit, an operationaware policy-learning framework that combines downstream outcome credit with immediate transition credit: Branch Outcome Credit (BOC) reuses ordinary rollouts to compare terminal task rewards observed after different operations under matched navigation contexts; Frontier Transition Credit (FTC) measures operation-induced changes in answer-conditioned frontier potential. The two complementary signals jointly supervise the corresponding decision spans during training, while inference uses only the learned policy and standard KG interface. Experiments on ComplexWebQuestions, WebQSP, and GrailQA across three backbones demonstrate consistent improvements in KGQA performance.

cs.CL↗
arXiv2026

MemoryAthena: Adaptive Routing over Latent and Generated Memories

Learned-memory methods store information in an explicit table and consume it through a separate reader, allowing addressing, storage, and reading to be modified independently. We study whether useful memory can also be generated rather than only retrieved. MemoryAthena uses three pathways: direct Engram retrieval (E), generation from retrieved Engram cues (GE), and generation from causal backbone states without consulting the memory table (GH). Generated memory is conditionally useful: it can complement E in one context but interfere with it in another. MemoryAthena therefore treats E as an anchor and learns when a generated representation should intervene. With the backbone, memory, generators, and readers frozen, a lightweight causal routing head is trained from counterfactual future-token likelihood advantages of GE and GH relative to E. At inference time, an admitted candidate modifies the E residual through bounded interpolation, while rejection recovers the direct pathway exactly. On question answering, MemoryAthena raises the five-task average from 37.65 to 39.28 over the direct pathway of the same checkpoint, while the six-task general-NLP average increases from 76.73 to 79.13. The complete memory-side system contains approximately 201M parameters, excluding the frozen backbone. Further analyses show complementary strengths among E, GE, and GH across tasks and inputs. These results support generated memory as a selective correction to direct retrieval and highlight routing when, which, and how strongly to intervene as the central challenge.

cs.CL↗
arXiv2026

Observation of stacked vortex structure with number controllable Majorana zero mode in Bi$_{2}$Te$_{3}$/FeTe heterojunction

Creation and control of Majorana zero modes (MZMs) in solid devices is the key challenge in realizing fault tolerant topological quantum computation. Here we report spectroscopic evidence for number controllable MZMs in stoichiometric one quintuple layer Bi$_{2}$Te$_{3}$/FeTe heterostructures, together with the observation of stacked vortices in superconducting islands. Low temperature scanning tunnelling microscopy and spectroscopy reveal a long period moire superlattice arising from lattice mismatch, which induces nanoscale modulation of superconducting gaps. Magnetic vortices exhibit a stacked core structure with two layer derived bound state branches: a dispersive branch evolving into an X-shaped pattern away from the vortex center, and a quantum-limit branch characterized by a spatially unsplit MZMs together with discrete integer quantized Caroli-de Gennes-Matricon states. Notably, MZMs are observed in all measured vortices, and their number can be tuned by magnetic field and the size of Bi$_{2}$Te$_{3}$ islands. These results establish stoichiometric 1QL-Bi$_{2}$Te$_{3}$/FeTe as a patterned superconducting interface that enables isolated and few-vortex Majorana configurations, providing a controlled platform for fuse MZMs.

cond-mat.supr-con↗
arXiv2026

Label-free steering: Compressing test-time reinforcement learning into bias-only subspaces

Test-time reinforcement learning (TTRL) enables models to improve their reasoning without relying on labeled training data, but existing approaches typically optimize a large fraction of the model parameters. This raises a natural question: can effective test-time adaptation emerge when both the reward signal and the optimization space are severely restricted? We answer this question with label-free bias-only TTRL, which uses majority-vote pseudolabels as rewards and optimizes only ~100K bias parameters while keeping the pretrained backbone frozen. On MATH-500, our approach reaches 76.67% accuracy with Qwen2.5-7B, slightly exceeding our own labeled bias-steering reproduction while optimizing 76,000x fewer parameters than full-parameter TTRL. The same training procedure improves performance across vision-language and audio reasoning tasks, including MathVista, AI2D, LogicVista, and MMAU. We further show that the learned steering vectors transfer to 4,500 held-out MATH problems, indicating that the adaptation is not limited to the problems used during test-time optimization. Finally, we analyze why this highly restricted adaptation can work, showing that majority-vote reliability improves with rollout consensus and that bias subspaces with greater accessible gradient energy exhibit stronger downstream trainability. These results demonstrate that substantial test-time adaptation can emerge from optimizing a tiny bias-only subspace using entirely label-free rewards.

cs.LG↗
arXiv2026

MDRC: A Deployable State-Recovery Defense for Traffic Signal Control under Sensor Corruption

Traffic Signal Control (TSC) is a safety-critical cyber-physical system that relies on real-time sensing. Corrupted observations caused by adversarial perturbations or sensor failures can propagate from the sensing layer into the controller and degrade traffic efficiency. Existing robust Reinforcement Learning (RL)-based TSC methods often suffer from limited cross-city generalization, high inference latency, and weak recovery under partial observability. We present MDRC (Meta-Diffusion-based framework for Resilient traffic signal Control against adversarial attacks and sensor failures), a post-detection state-recovery defense inserted between sensing and control. MDRC reconstructs trustworthy traffic states before they are consumed by the controller. It combines Denoising Diffusion Implicit Models (DDIM) for efficient state recovery with Reptile meta-learning for a transferable initialization across cities. We provide an optimization-based view of the DDIM recovery dynamics and establish a recovery-error bound that separates score approximation, numerical discretization, and initialization mismatch. Across seven real-world-derived CityFlow benchmarks, MDRC reduces Average Travel Time by 6.77% under stochastic and policy-aware attacks and by 12.75% under structured sensor loss, while improving state-recovery fidelity. We further evaluate 3,600 seconds of real roadside measurements with 50% of detector channels disabled and integrate MDRC into a hardware-in-the-loop traffic-signal stack. Over a 9.16-hour run with 32,389 sensing/control cycles, the system achieves 99.79% decision availability, produces no out-of-plan recommendations, and requires approximately 38 ms of component-wise processing per one-second control interval.

cs.CR↗
arXiv2026

Your Model Is Leaking: Covert Information Transfer through LLM Residual Streams

Privacy-sensitive organizations may run large language models (LLMs) in restricted or air-gapped environments while exporting selected diagnostic artifacts. We show that a compromised runtime component can hide sensitive information in intermediate activations that are allowed to leave the restricted environment. An offline observer can recover this information with a simple linear decoder. The attack requires no model retraining or weight modification, no attacker-controlled egress, and no control over the recorder or transfer process. We introduce a residual-stream covert-channel attack that maps messages to codewords and injects them into an intermediate residual stream through a compromised runtime hook. To maintain recoverability, the injection strength is scaled with the local residual norm using the signal-to-residual-norm ratio. Across eleven models from seven architecture families, our evaluation shows 91--100% recovery on nine models with KL divergence 0.001--0.007, while evaluated activation-level detectors remain close to random guessing (AUC <= 0.56). Tested post-hoc defenses do not reliably eliminate the channel. Thus, an activation artifact can be schema-valid while carrying information that is not authorized to cross the boundary.

cs.CR↗
arXiv2026

A High-Payload Wall-Climbing Robot Using Passive Bistable Suction Cups

Wall-climbing robots capable of scaling vertical surfaces could help automate hazardous or labor intensive tasks such as window washing, inspection, maintenance, and construction. Active adhesion methods achieve higher payload capacities, but require power to maintain their grip. Passive adhesion devices such as suction cups are an attractive option for such robots because they do not require power to maintain their grip, but they are limited by their payload capacity. This work presents a novel high-payload wall-climbing robot that utilizes passive bistable suction cups to generate adhesion without needing to be pushed into the wall. The robot features a track-based system that automatically engages and disengages bistable suction cups to achieve locomotion on smooth surfaces. The robot is able to achieve vertical wall climbing on glass, wood, metal, and painted surfaces, sideways and upside-down climbing, and is able to tow a payload of 7.940 kg (with a payload-to-weight ratio of 2.25).

cs.RO↗
arXiv2026

CacheWeaver: Cache-Aware Evidence Ordering for Efficient Grounded RAG Inference

Retrieval-Augmented Generation (RAG) improves factual grounding, but it also lengthens prompts and raises prefill cost. Prefix caching in serving engines such as vLLM reduces this cost only when requests share the same token prefix. In grounded generation, however, adjacent queries may retrieve overlapping evidence in different orders, so set overlap does not become reusable prefix overlap. We present CacheWeaver, a lightweight prompt-layer method for cache-aware evidence ordering. The method keeps a prefix tree over recently served evidence sequences and uses a greedy walk to place the most reusable prefix first, while leaving the serving engine and retrieved evidence set unchanged. Across three vLLM configurations, the method lowers median time-to-first-token (TTFT) by about 20-33 percent relative to retrieval-order prefix caching, without hurting answer quality in our QA tests. The greedy policy reaches 97.5 percent of the median TTFT gain from oracle ordering, indicating that most reusable prefix locality can be recovered by a simple scheduling layer between retrieval and inference.

cs.CL↗
arXiv2026

Test Case Prioritization for DNNs via Neural Collapse Instability

With the widespread deployment of deep neural networks (DNNs) in safety-critical domains, reducing the cost of model validation under limited testing budgets has become increasingly important. Existing test case prioritization techniques often rely on single-checkpoint confidence signals derived from output probabilities. However, DNNs can be confidently wrong, and the confidence margin between the predicted and competing classes is frequently small, which weakens early fault discovery. To address this limitation, we propose a Neural-Collapse-Inspired Prioritization (NCIP) framework that replaces absolute confidence with cross-checkpoint prediction variability in the terminal training regime, where model geometry becomes highly structured. NCIP introduces two key components. First, it selects an NC-guided representative subset of training checkpoints using an equiangularity score of classifier weights, quantified as the standard deviation of pairwise cosine similarities among class weight vectors. Second, it prioritizes test inputs by their prediction variability across the selected checkpoints, surfacing boundary-adjacent and failure-prone samples that are unstable under checkpoint-induced decision boundary shifts. Extensive experiments across multiple datasets and architectures show that NCIP achieves strong performance in early fault discovery compared with competitive baselines, with 1.5 to 16.6 percent RAUC-ALL gains and 4.9 to 20.6 percent RAUC-500 gains under the same testing budget. NCIP further attains the best average performance across all dataset-model pairs.

cs.LG↗
arXiv2026

A Collocated Boris Integrator in Flux Coordinates: Balancing Accuracy, Conservation, Cost and Robustness

When the guiding-center description fails and the full gyromotion must be resolved for energetic particles in complex configurations like stellarators, charged-particle integrators must be formulated directly in the curvilinear flux coordinates. The Boris algorithm, which adopts a staggered scheme in Cartesian coordinates, is phase-space-volume-preserving and second-order accurate; but a direct port to flux coordinates degrades the position update to first order, because the evolving basis vectors of the curvilinear frame make the starting-point metric deviate from the ideal midpoint metric. We construct a collocated, midpoint-predicted Boris algorithm in flux coordinates, restoring second-order accuracy at the cost of one additional field evaluation per step. In reactor-scale stellarator magnetic fields, the scheme recovers second-order convergence in every coordinate component, retains near-machine-precision energy conservation and a bounded magnetic moment, and demonstrates greater orbit robustness than Staggered Boris and RK4 at coarse time steps.

physics.plasm-ph↗
arXiv2026

Real-Time Compliance and Position Control of a Hyper-redundant Soft Robotic Arm

Robots working in unstructured or partially unobservable environments must combine accurate motion with physical compliance that can passively correct contact misalignment. Soft robots provide this compliance but have struggled to precisely control their tip compliance and position. This paper presents a robot architecture designed around that control problem: a 7-link arm whose six articulated joints provide twelve independently driven revolute axes, each actuated by an antagonistic pair of pneumatic muscles, so that every axis can simultaneously change its angle and linearly adjust its stiffness. The rigid articulated backbone makes the tip compliance and position of the arm predictable enough to be commanded quantitatively in real time. The robot employs a unified iterative inverse-kinematics and inverse-compliance controller to achieve simultaneous, quantitative control of both compliance and position. The task-space compliance and kinematics models and the control law are derived and verified on both the physical arm and a matched simulation. Simulation is then used to study how the same framework extends to other arm morphologies. Finally, the arm demonstrates tasks that have been difficult for both rigid and soft arms: rejecting disturbances while writing on a moving whiteboard, and passively correcting hidden misalignment during a key-insertion and drawer-opening task. That these tasks succeed under so straightforward a controller is evidence for the advantage of this algorithm-informed structural design.

cs.RO↗
arXiv2026

OverFlowLight: Real-Time Gridlock Prevention and Traffic Signal Optimization for Urban Intersections

Queue overflow, a severe consequence of urban traffic congestion, occurs when vehicle queues exceed intersection capacity, obstructing upstream traffic and triggering cascading gridlocks. Prevailing traffic signal control (TSC) algorithms, primarily optimized for throughput, often fail to address overflow during peak hours, exacerbating congestion and creating safety hazards. We propose OverFlowLight, a real-time framework designed to preemptively resolve overflow and enhance overall TSC performance. It first introduces a mechanism to accurately detect overflow in real-time by leveraging multi-modal sensing from cameras and radars. Upon detection, it dynamically generates and inserts dedicated overflow phases into the signal cycle to clear the blocking queues. This is orchestrated by a hybrid control design that combines rapid rule-based overflow intervention with controller back ends such as reinforcement learning (RL) for longer-horizon efficiency. We conducted extensive real-world deployments of OverFlowLight across 43 intersections in three major cities. The framework demonstrates seamless integration with existing RL-based TSC agents, highlighting its modularity and practical applicability. Empirical results show that OverFlowLight reduces overflow incidents by 60.4% and increases network throughput by 18.2% compared to deployed baselines. Furthermore, it substantially diminishes the need for manual intervention common with expert-tuned signal plans. This work presents the first practical, scalable, and data-driven framework for actively preventing traffic gridlock, offering a crucial component for building resilient and efficient urban transportation systems. Our demonstration videos, codes and datasets are available at the anonymous URL, https://anonymous.4open.science/r/OverFlowLight-FBF9.

cs.LG↗