arXiv subjects

Zhen Liu

Publications and source records attributed to Zhen Liu.

At least 19 recordsLinked to original sources

arXiv2026

$S^3$-Bench: Evaluating Speech Interaction Models as Scientific Voice Assistants

The advance of multimodal large language models (MLLMs) has fundamentally reshaped the paradigm of human-computer interaction, especially speech interaction models capable of seamless conversations. Despite remarkable performance as general voice assistants, their performance in specialized domains remains underexplored, particularly in scientific areas. Scientific interactions introduce formidable challenges, involving rare technical terminology, spoken norms of abbreviations, and the natural verbalization of symbolic special expressions. In this paper, we introduce S$^3$-Bench, a systematic evaluation framework covering 10 major disciplines, consisting of a Knowledge set for speech question-answering and a Dialogue set for multi-turn progressive interactions with simulated user agents. By decomposing a complete atomic turn into stages of speech recognition, perception, knowledge utilization with reasoning, and response pronunciation, we systematically characterize the common challenges and performance tradeoffs of existing approaches. Furthermore, experiments on multi-turn interactions reveal persistent limitations in user adaptation and the generation of accurate, comprehensive, and efficient responses.

cs.CL↗

arXiv2026

Under-Mattress Temporal Sensing for Next-Day Agitation Risk Scoring in Dementia Wards

Agitation fluctuates over short time horizons in people living with dementia, yet continuous physiological information for anticipating next-day risk is limited. We assessed whether contactless under-mattress signals from the preceding night inform next-day agitation risk and whether preserving minute-level temporal structure improves performance over conventional nightly summaries. We analyzed 423 patient-nights from 65 subjects in a specialized hospital dementia unit using two under-mattress sensing systems. A unified four-paradigm benchmark compared nightly handcrafted summaries, three-period handcrafted features, full-night sequence modeling, and sliding-window multiple-instance learning. Source-specific preprocessing and five-fold patient-grouped cross-validation were used, with performance estimated from pooled out-of-fold predictions. Evaluation included discrimination, calibration, fixed-threshold metrics, and a comparison of period-signal attribution patterns across two temporal models. Full-night sequence modeling achieved the highest discrimination (AUROC, 0.692; AUPRC, 0.849) and balanced accuracy (0.658). Both minute-level pipelines had higher AUROC than nightly summaries, but differences from three-period handcrafted features were uncertain. Cross-model attribution prioritized activity, heart rate, and respiratory rate during the core overnight period. Calibration remained limited. The preceding night's signals supported modest next-day risk discrimination, with minute-level temporal modeling outperforming nightly summaries. Prospective calibration and external validation are needed before use in individual care decisions. This patient-grouped benchmark identifies contactless overnight sensing as a promising biomedical engineering direction for agitation-risk research in hospitalized dementia cohorts.

cs.AI↗

arXiv2026

Contrastive Representation-Guided Genetic Minority Oversampling for Imbalanced Time-Series Classification

Real-world time-series classification tasks often exhibit class imbalance, which can be extremely severe in some applications. To avoid training biased classifiers on imbalanced data, sampling is one of the most popular data pre-processing techniques because of its classifier-agnostic nature. However, due to the complex temporal dependencies in original time-series data and the scarcity of minority-class samples, existing sampling methods, including interpolation-based oversampling methods and deep learning-based generative models, usually suffer from limited generalization and poor diversity when generating new time-series samples. This paper proposes a Frequency-domain representation-guided Multi-tree Genetic Programming-based oversampling approach (FreMGP) to imbalanced time-series classification, where each individual represents a set of synthetic samples for the minority class. A frequency-domain class-discriminative representation module based on contrastive learning is also developed, guiding the evolutionary search toward high-quality synthetic time-series samples. Experiments on imbalanced time-series datasets demonstrate that FreMGP outperforms existing oversampling methods and consistently improves the performance of different classifiers, including both general machine learning and deep learning models.

cs.LG↗

arXiv2026

A subquadratic bound for generalized Tur\'an numbers of odd cycles

For a graph $H$ and a family of graphs $\mathcal F$, let $\text{ex}(n,H,\mathcal F)$ denote the maximum number of copies of $H$ in an $\mathcal F$-free graph on $n$ vertices. For every integer $i\ge 3$, let $C_i$ denote the cycle of length $i$. For $r\ge 3$, set $\mathscr {C}_r=\{C_3,C_4,\ldots,C_r\},$ and set $\mathscr {C}_2=\varnothing$. In this paper, we prove that, for all integers $l>k\ge 2$, $$ \text{ex}(n,C_{2k+1},\mathscr {C}_{2k}\cup\{C_{2l+1}\}) =O_{k,l} \left(n^{2-\frac{1}{k(k+1)(l-k)}}\ \ \right). $$ Together with the known upper bounds for the number of triangles in $C_{2l+1}$-free graphs, this confirms a conjecture of Gerbner, Gy\H{o}ri, Methuku, and Vizer.

math.CO↗

arXiv2026

Sharp asymptotics for triangle independence and covering numbers

For a graph $G$, let $\alpha_1(G)$ be the maximum size of an edge set containing at most one edge from every triangle, and let $\tau_1(G)$ be the minimum size of an edge set meeting every triangle. Erd\H{o}s, Gallai, and Tuza proved that $\alpha_1(G)+\tau_1(G)=\Omega(m^{2/3})$ for every $m$-edge graph and asked for the optimal asymptotic constant. We prove $$\lim_{m\to\infty} \min_{G,\,|E(G)|=m} \frac{\alpha_1(G) + \tau_1(G)}{m^{2/3}} = \frac{3}{2},$$ thereby establishing that the sharp constant is $3/2$ and solving the problem.

math.CO↗

arXiv2026

Mean Field Stackelberg Game for Production and Carbon Emission Reduction with State Reflections

Global warming, driven by anthropogenic carbon emissions with transboundary pollution characteristics and irreversible damage, poses an existential threat to human society. This paper develops a novel two-level Stackelberg game with mean field interaction of controls and common noise which integrates hierarchical decision-making under a state-reflected emission dynamics. A central regulator (leader) adjusts product prices to guide $n$ heterogeneous competing regions (followers) while enforcing a hard emission cap via a reflection mechanism that models emergency reductions through a local time process. We establish the existence of an approximate Stackelberg equilibrium and perform sensitivity analysis via Monte Carlo simulations.

math.OC↗

arXiv2026

A sharp asymptotic bound for odd cycles in planar graphs

For graphs $G$ and $H$, let $\mathbf N(G,H)$ denote the number of unlabeled, not necessarily induced copies of $H$ in $G$, and let $\mathbf N_{\mathcal P}(n,H)$ be the maximum of $\mathbf N(G,H)$ over all $n$-vertex planar graphs $G$. We prove that, for every fixed integer $m\geq 3$, $$\mathbf N_{\mathcal P}(n,C_{2m+1})=2m\left(\frac{n}{m}\right)^m+O_m\!\left(n^{m-1/5}\right).$$ The proof uses a sharp weighted cycle--path inequality for edge probability measures on finite complete graphs. This strengthens a conjecture of Heath, Martin, and Wells and, together with their reduction lemma, yields the stated asymptotic formula.

math.CO↗

arXiv2026

An AI4AI Framework for Visual Token Pruning

Visual-token pruning can substantially reduce the inference cost of multimodal large language models (MLLMs), yet existing methods largely rely on fixed, handcrafted heuristics and costly expert trial and error. As pruning objectives, budgets, and model architectures diversify, manually navigating the expanding design space becomes increasingly difficult. This paper aims to build an AI4AI framework for visual-token pruning by addressing a natural question: Can large language models automatically design effective visual-token reduction algorithms? Although LLMs possess broad algorithmic knowledge and strong reasoning capabilities, translating such general knowledge into effective solutions for a specialized task remains nontrivial. We argue that the key lies in designing an appropriate search-state representation that connects the internal knowledge of LLMs with the structural requirements and constraints of visual-token pruning. Based on this insight, we propose AutoPrune, a training-free framework for LLM-driven visual-token pruning policy design. At its core, AutoPrune introduces a Token Pruning Domain-Specific Language (TPDSL) comprising 131 reusable atoms for budget control, token scoring, selection constraints, and token reassembly. A key property of TPDSL is that it represents each search state as a residual modification of a strong base policy. This residual formulation narrows the search space and directs the LLM's attention toward the policy components that are most consequential for performance. Experiments on 14 multimodal benchmarks and three MLLM backbones demonstrate the effectiveness, efficiency, and transferability of AutoPrune. Even when removing 94.4% of visual tokens, AutoPrune preserves more than 99% of full-token performance while reducing FLOPs by 9.9x and prefill latency by 6.4x.

cs.LG↗

arXiv2026

GraphIR: Architecture-Level Search States for LLM-Guided Neural Architecture Evolution

Large language models (LLMs) enable neural architecture search (NAS) directly over executable neural network programs. However, code-level flexibility does not provide the architecture state needed for effective mutation: LLMs must infer tensor dependencies, editable components, and compatibility constraints from implementation details. To address this representation mismatch, we propose GraphIR, an architecture-aware intermediate representation that supplements executable programs with a mutation-aligned candidate state. GraphIR organizes each candidate through three complementary views: a computation skeleton describing tensor flow, a mutation surface exposing editable modules and operations, and a validity envelope capturing interface contracts, propagated shapes, and downstream dependencies. To evaluate our method, we construct NAS-Dependency, a 120-question benchmark covering six complementary dependency-reasoning dimensions. The diagnostic shows that GraphIR is particularly effective at identifying exact producer occurrences, tracing dependency propagation, and diagnosing interface and failure risks. Across six downstream benchmarks including CLRS, GraphIR achieves the best overall search performance while maintaining comparable model size and favorable end-to-end NAS efficiency when integrated into OpenEvolve. These results show that a mutation-oriented architecture state provides an effective interface between executable neural programs and LLM-guided architecture evolution.

cs.LG↗

arXiv2026

The Forward Neutrino Flux and its Secondaries at a 10 TeV Muon Collider

Muon decays in a muon collider ring would produce TeV neutrino and antineutrino beams of electron and muon flavor. We study this flux in the forward $\mu^+$ and $\mu^-$ beam directions at a 10 TeV muon collider and introduce \mint, a dedicated Monte Carlo simulation to model neutrino fluxes accounting for muon beam dynamics. We find that a benchmark detector at 5 km from the interaction point would see about $\mathcal{O}(10^{9})$ neutrino interactions per year in a $\sim3$~tonne fiducial volume with a beam spot size of $\mathcal{O}(1)$~meter. We calculate the number of secondary muons and neutrinos generated by neutrino interactions in the rock upstream of the forward detectors and find that about two secondary high-energy and highly polarized muons from the rock would cross each detector per bunch crossing. Neutrino productions of charmed mesons and taus in the rock generate a small $\nu_\tau+\bar\nu_\tau$ secondary flux, with $\mathcal{O}(0.2)$ events per year in the detectors, likely too small to be observed. Wrong-sign neutrinos from similar processes, such as $\nu_e+\bar\nu_\mu$ in the $\mu^-$ beam, are more numerous but still a fraction no larger than $\mathcal{O}(10^{-8})$ of the number of TeV neutrino interactions. Finally, we outline how the large forward neutrino exposure can be used to search for beyond-the-Standard-Model particles produced in neutrino interactions, with examples of heavy neutral leptons coupled to electron and muon flavors through mixing or electromagnetic dipole operators.

hep-ph↗

arXiv2026

Overcoming the Weakest-Link Effect in LLM-Driven Program Optimization via Heterogeneous Edit Recombination

Large language models (LLMs) are increasingly used to solve complex problems by searching over program space, offering a general paradigm for scientific problems that can be naturally represented and solved as programs. Despite recent progress, identifying effective optimization directions for a candidate program remains challenging. By analogy with automatic differentiation, existing methods typically guide the search using a textual ``gradient'': a first-order update direction expressed as textual edits. Such gradients are inferred either from previously evaluated programs or from LLM-generated feedback on the implicit program-score mapping. However, these estimates become increasingly unreliable as the program--score mapping grows more complex, limiting their practical utility. We argue that explicit gradients are not essential for effective program optimization. Leveraging their prior knowledge, LLMs can propose plausible atomic edits directly from the current program, thereby enabling a zeroth-order optimization strategy. However, zeroth-order search suffers from a \textit{weakest-link effect}: when a bundle of edits is accepted or rejected as a whole, a single harmful edit can negate the benefits of all remaining edits. To address this issue, we introduce HERO, a program optimizer that prompts an LLM to generate diverse, non-overlapping atomic edits and then systematically selects and composes them into coherent program improvements using evaluator scores. We evaluate HERO across algorithmic problems, strategy games, the design of LLM-based agentic systems, and robotic path planning. Across these domains, HERO consistently discovers higher-scoring programs and converges substantially faster than prior LLM-based optimizers, while consuming fewer tokens.

cs.LG↗

arXiv2026

The maximum number of paths of even length in a planar graph

For graphs $G$ and $H$, let $N(G,H)$ be the number of unlabeled, not necessarily induced copies of $H$ in $G$, and let $f(n,H)$ be the maximum of $N(G,H)$ over all $n$-vertex planar graphs $G$. Ghosh, Gy\H{o}ri, Martin, Paulos, Salia, Xiao and Zamora conjectured that, for every fixed integer $\ell\ge 2$, \[ f(n,P_{2\ell+1}) =4\ell\left(\frac{n}{\ell}\right)^{\ell+1}+O(n^\ell). \] We prove the conjecture, including the stated error term. Along the way, we also settle the Cox--Martin optimization conjecture.

math.CO↗

arXiv2026

Extreme Volatility Warning under Label Scarcity via Multi-Source Anomaly Fusion

Early warning of extreme market volatility is central to financial risk management, but actionable events are rare, nonstationary, and often triggered by exogenous information shocks. In our CSI~300 setting, only $\sim$80 positive samples are observed across 791 training days, making heavily supervised multi-source models unstable. We first analyze a 100K-parameter hierarchical text-signal fusion model (HTSF) and find that added parameterization hurts in this low-label regime. Motivated by this failure, we propose \textbf{AAMSF} (Anomaly-Augmented Multi-Signal Fusion), a semisupervised framework that combines Isolation Forest anomaly scores over market indicators, GDELT events, Chinese financial news, and English media with lightweight Ridge score fusion. We further introduce \textbf{T-AAMSF}, a temporal extension for multi-day anomaly accumulation. On CSI~300 (2018--2023), AAMSF achieves test AUC-ROC \textbf{0.680}, outperforming the strongest unsupervised baseline (0.630) and neural baseline (0.588), while T-AAMSF improves PR-AUC to 0.291. Ablations reveal strong source asymmetry: GDELT and domestic financial news provide complementary risk signals, whereas English media consistently reduces performance, and learned weighting is unreliable under validation noise. These results suggest an empirical design principle for label-scarce financial risk warning: robust anomaly geometry and source reliability can matter more than supervised representation capacity.

cs.LG↗

arXiv2026

SARATR-X-v2: Scale-Aware Structural Pre-Training for SAR Foundation Models

Masked image modeling has become a dominant paradigm for SAR pre-training, yet the design of the reconstruction target remains fundamentally unsettled. This article argues that a SAR pre-training target should satisfy two conditions to produce transferable representations: (i) physics-grounded stability, i.e., approximate invariance of the target operator to multiplicative speckle inherent in coherent imaging; and (ii) semantic scale compatibility, i.e., coverage of the heterogeneous spatial scales that downstream tasks demand. These two conditions are individually achievable but jointly difficult: physics-grounded stability favors fixed operators, while semantic scale compatibility favors data-driven composition. To this end, SARATR-X-v2 reconciles both within a single design. The target is constructed through fixed structural extractors spanning six receptive fields, from blind-spot local aggregation to directional log-ratio region contrast, and fused via learnable weights into one unified supervision signal for masked reconstruction. On twelve SAR benchmarks across classification, detection, and segmentation, SARATR-X-v2 achieves state-of-the-art transfer performance. Under synthetic speckle variation, the proposed target reduces perturbation drift in the learned representation by nearly two orders of magnitude relative to pixel-space supervision. Taken together, these results establish physics-grounded stability and semantic scale compatibility as a principled framework for pre-training target design under coherent imaging, and suggest that effective SAR pre-training is not about reconstructing more signal, but about reconstructing the right structural target.

cs.CV↗

arXiv2026

Toward Active Object Detection for UAVs in the Wild: A Large-Scale Dataset, Benchmark and Method

Object detection is a fundamental component in numerous Unmanned Aerial Vehicle (UAV) applications, yet it has long been plagued by hindrances like occlusion or target pixel scarcity. Active Object Detection (AOD) provides a novel paradigm to address these challenges via active vision, while UAV-based AOD research remains scarce due to the lack of high-quality datasets and benchmarks for algorithm development and evaluation. To fill this gap, this paper presents ATRNet-LUDO, the first large-scale real-world dataset for UAV-Ground Active Object Detection (UGAOD). It contains 121,000 multi-view panoramic multi-target aerial images and 1.21 million local single-target slices, covering 10 vehicle targets across 40 scenarios. It enables the construction of diverse training and testing environments for UAV agent interaction and active observation policy learning. Based on this dataset, we establish a comprehensive evaluation benchmark for AOD policy learning methods. Most existing AOD policies rely on Deep Reinforcement Learning (DRL) but suffer from poor generalization. Evaluations on our benchmark reveal a significant generalization gap between training and testing performance, highlighting an urgent need for solutions. To this end, we leverage the Joint Embedding Predictive Architecture (JEPA) to construct a world model that enhances state representation learning, and propose AOD-JEPA by incorporating AOD-specific prior knowledge. Extensive experiments validate its effectiveness and superiority. We hope ATRNet-LUDO and the benchmark will advance research in the UGAOD field. The dataset and code are soon available at https://github.com/Leo000ooo/LUDO_dataset.

cs.CV↗

arXiv2026

Paths of even length with equal-degree endpoints

Addressing a question posed by Erd\H{o}s and Hajnal, Chen and Ma proved that, for all $n \ge 600$, the complete bipartite graph $K_{n,n+1}$ is the unique graph on $2n+1$ vertices with at least $n^2+n$ edges that contains no two vertices of equal degree joined by a path of length three. In this paper, we extend this result and prove that for every fixed integer $\ell\ge 2$ and sufficiently large $n$, the unique $2n$-vertex graph with at least $(n^2+n)/2$ edges that contains no two vertices of equal degree joined by a path of length $2\ell$ is the half graph $H_n$. This resolves the problem posed by Chen and Ma, as well as a related question of Attwa, Az\'ocar Carvajal, Boyadzhiyska, Pierron, and Taraz concerning paths of even length with equal-degree endpoints.

math.CO↗

arXiv2026

DIPBox: A Multi-scale Testing Framework for Tracking Dataset Regeneration

Training datasets have tremendous proprietary value and are vulnerable to unauthorized copying. Existing defenses mainly focus on tracking individual data points, but pay little attention to the threat of dataset regeneration. Through a measurement study of public tumor datasets, we identify substantial real-world partial-dataset replication, raising concerns about potential license noncompliance. To counter the challenge of tracking previously unknown adversarial regeneration, our key insight is that regeneration that preserves model utility inevitably preserves measurable signals across multiple feature scales. We categorize these dataset features into sample-, set-, and distribution-level features and design four similarity metrics to accurately identify regeneration. Based on these metrics, we develop DIPBox, which to our knowledge is the first testing framework that tracks regeneration suspects via multi-scale similarity testing across a spectrum of defender access settings, from limited to full information. We further provide a learning-theoretic analysis that justifies these multi-scale metrics and formalizes an inherent utility--divergence trade-off, implying fundamental limits on evasive regeneration. Extensive experiments on 16 vision and text base datasets, 320 regenerated datasets, and 590 derived models validate that DIPBox outperforms previous solutions while characterizing its robustness and limits under three adaptive attacks.

cs.CR↗

arXiv2026

ELVA: Exploring Ranking-Driven Universal Multimodal Retrieval

Leveraging Multimodal Large Language Models (MLLMs) via contrastive learning has become a mainstream paradigm for improving the performance of Universal Multimodal Retrieval (UMR). However, previous works have ignored the grain blindness when adapting the contrastive paradigm into retrieval tasks. Grain blindness refers to the tendency of the model to overlook grain-level information contained in the query, which is crucial for effectively handling complex queries. This stems from contrastive learning treating samples as a binary classification (positive/negative), while ignoring the different information carried by each negative sample. To address this, we argue that negatives should be treated differently according to their similarity to the positive sample, enabling the model to learn distinct grain information from each negative. In this paper, we introduce a simple but effective framework, called ELVA, a novel rule-based RL framework that mitigates grain blindness through ranking-driven MLLMs. 1) Instead of relying on reward models, we extend Reinforcement Learning with Verifiable Rewards (RLVR) to retrieval tasks, allowing the model to explore new ranking behaviors without explicit ranking labels. 2) By utilizing rule-based rewards, our approach jointly optimizes the ranking of negative samples while enlarging the similarity gap between positive and negative. To more precisely measure grain blindness, we further introduce MRBench, a new benchmark specifically designed for multi-grain query scenarios. ELVA achieves state-of-the-art results across standard retrieval benchmarks, and its notable 13.1% improvement on MRBench further demonstrates its effectiveness in alleviating grain blindness.

cs.IR↗