arXiv subjects

Tao Zhang

Publications and source records attributed to Tao Zhang.

At least 181 records · Page 10Linked to original sources

arXiv2026

Integrated Sensing, Communication and Control enabled Agile UAV Swarm

Uncrewed aerial vehicle (UAV) swarms are pivotal in the applications such as disaster relief, aerial base station (BS) and logistics transportation. These scenarios require the capabilities in accurate sensing, efficient communication and flexible control for real-time and reliable task execution. However, sensing, communication and control are studied independently in traditional research, which limits the overall performance of UAV swarms. To overcome this disadvantage, we propose a deeply coupled scheme of integrated sensing, communication and control (ISCC) for UAV swarms, which is a systemic paradigm that transcends traditional isolated designs of sensing, communication and control by establishing a tightly-coupled closed-loop through the co-optimization of sensing, communication and control. In this article, we firstly analyze the requirements of scenarios and key performance metrics. Subsequently, the enabling technologies are proposed, including communication-and-control-enhanced sensing, sensing-and-control-enhanced communication, and sensing-and-communication-enhanced control. Simulation results validate the performance of the proposed ISCC framework, demonstrating its application potential in the future.

eess.SP↗

arXiv2026

Learning to Dock: Geometric Deep Learning for Predicting Supramolecular Host-Guest Complexes

Predicting non-covalent host-guest recognition remains challenging due to the complex interplay of electrostatics, dispersion, and steric effects, and the limited transferability of existing docking approaches to synthetic supramolecular systems. Here we present DeepHostGuest, a geometric deep-learning framework that learns generalizable recognition principles directly from experimentally resolved host-guest structures. Hosts are encoded as electrostatic surfaces and guests as molecular graphs, enabling transferable learning across diverse supramolecular systems. DeepHostGuest achieves high-accuracy predictions (RMSD $\leq 2$ Angstrom for 80.8% of test cases), substantially outperforming classical docking without case-specific tuning. Notably, the method generalizes beyond its training domain to crystalline sponge systems, accurately capturing the binding of large amphiphilic molecules within metal-organic cages. Beyond predicting binding conformations, the structures generated by DeepHostGuest serve as a reliable basis for accurate binding free-energy calculations. Density Functional Theory (DFT)-calculated affinities correlate well with experiment, enabling structure-property relationships across 876 host-guest complexes spanning 34 host families. Interpretable feature analysis reveals that binding strength arises from a cooperative interplay of host polarity, guest hydrophobicity, and geometric complementarity, with distinct design regimes across supramolecular classes. Together, these results establish data-driven molecular recognition as a practical route to predictive supramolecular design, enabling high-throughput virtual screening and rational optimization of functional host-guest systems.

physics.chem-ph↗

arXiv2026

Differentially Private Subspace Fine-Tuning for Large Language Models

Fine-tuning large language models on downstream tasks is crucial for realizing their cross-domain potential but often relies on sensitive data, raising privacy concerns. Differential privacy (DP) offers rigorous privacy guarantees and has been widely adopted in fine-tuning; however, naively injecting noise across the high-dimensional parameter space creates perturbations with large norms, degrading performance and destabilizing training. To address this issue, we propose DP-SFT, a two-stage subspace fine-tuning method that substantially reduces noise magnitude while preserving formal DP guarantees. Our intuition is that, during fine-tuning, significant parameter updates lie within a low-dimensional, task-specific subspace, while other directions change minimally. Hence, we only inject DP noise into this subspace to protect privacy without perturbing irrelevant parameters. In phase one, we identify the subspace by analyzing principal gradient directions to capture task-specific update signals. In phase two, we project full gradients onto this subspace, add DP noise, and map the perturbed gradients back to the original parameter space for model updates, markedly lowering noise impact. Experiments on multiple datasets demonstrate that DP-SFT enhances accuracy and stability under rigorous DP constraints, accelerates convergence, and achieves substantial gains over DP fine-tuning baselines.

cs.LG↗

arXiv2026

GPS-Synchronized Monitoring of Core-collapse Supernova Bursts with PandaX-4T via Coherent Elastic Neutrino Nuclear Scattering

The landmark detection of neutrinos from SN1987A marked the dawn of neutrino astrophysics. The neutrino burst provided essential insights into fundamental properties of neutrinos, and served as key probes of stellar evolution and supernova dynamics. The recent advancement in coherent elastic neutrino-nucleus scattering enables the detection of core-collapse supernova burst neutrinos using tonne-scale liquid xenon detectors originally designed for dark matter direct detection. Leveraging this capability, we developed and deployed an online supernova monitoring system for the PandaX-4T experiment. This system features a GPS module with millisecond-level timing precision, a low false-alarm rate, and high sensitivity to galactic core-collapse supernova explosion events. The methodology is robust, directly scalable, and planned for implementation in the next-generation PandaX-20T experiment.

hep-ex↗

arXiv2026

Residual-PAC Privacy: Automatic Privacy Control Beyond the Gaussian Barrier

The Probably Approximately Correct (PAC) Privacy framework [46] provides a powerful instance-based methodology to preserve privacy in complex data-driven systems. Existing PAC Privacy algorithms (we call them Auto-PAC) rely on a Gaussian mutual information upper bound. However, we show that the upper bound obtained by these algorithms is tight if and only if the perturbed mechanism output is jointly Gaussian with independent Gaussian noise. We propose two approaches for addressing this issue. First, we introduce two tractable post-processing methods for Auto-PAC, based on Donsker-Varadhan representation and sliced Wasserstein distances. However, the result still leaves wasted privacy budget. To address this issue more fundamentally, we introduce Residual-PAC (R-PAC) Privacy, an f-divergence-based measure to quantify privacy that remains after adversarial inference. To implement R-PAC Privacy in practice, we propose a Stackelberg Residual-PAC (SR-PAC) privatization mechanism, a game-theoretic framework that selects optimal noise distributions through convex bilevel optimization. Our approach achieves efficient privacy budget utilization for arbitrary data distributions and naturally composes when multiple mechanisms access the dataset. Our experiments demonstrate that SR-PAC consistently obtains a better privacy-utility tradeoff than both PAC and differential privacy baselines.

cs.CR↗

arXiv2025

TableGPT-R1: Advancing Tabular Reasoning Through Reinforcement Learning

Tabular data serves as the backbone of modern data analysis and scientific research. While Large Language Models (LLMs) fine-tuned via Supervised Fine-Tuning (SFT) have significantly improved natural language interaction with such structured data, they often fall short in handling the complex, multi-step reasoning and robust code execution required for real-world table tasks. Reinforcement Learning (RL) offers a promising avenue to enhance these capabilities, yet its application in the tabular domain faces three critical hurdles: the scarcity of high-quality agentic trajectories with closed-loop code execution and environment feedback on diverse table structures, the extreme heterogeneity of feedback signals ranging from rigid SQL execution to open-ended data interpretation, and the risk of catastrophic forgetting of general knowledge during vertical specialization. To overcome these challenges and unlock advanced reasoning on complex tables, we introduce \textbf{TableGPT-R1}, a specialized tabular model built on a systematic RL framework. Our approach integrates a comprehensive data engineering pipeline that synthesizes difficulty-stratified agentic trajectories for both supervised alignment and RL rollouts, a task-adaptive reward system that combines rule-based verification with a criteria-injected reward model and incorporates process-level step reward shaping with behavioral regularization, and a multi-stage training framework that progressively stabilizes reasoning before specializing in table-specific tasks. Extensive evaluations demonstrate that TableGPT-R1 achieves state-of-the-art performance on authoritative benchmarks, significantly outperforming baseline models while retaining robust general capabilities. Our model is available at https://huggingface.co/tablegpt/TableGPT-R1.

cs.LG↗

arXiv2025

Information-Scrambling-Enhanced Quantum Sensing Beyond the Standard Quantum Limit

Quantum sensing promises measurement precision beyond classical limits, but its practical realization is often hindered by decoherence and the challenges of generating and stabilizing entanglement in large-scale systems. Here, we experimentally demonstrate a scalable, scrambling-enhanced quantum sensing protocol, referred to as butterfly metrology, implemented on a cross-shaped superconducting quantum processor. By harnessing quantum information scrambling, the protocol converts local interactions into delocalized metrologically useful correlations, enabling robust signal amplification through interference of the scrambled and polarized quantum states. We validate the time-reversal ability via Loschmidt echo measurements and quantify the information scrambling through out-of-time-ordered correlators, establishing the essential quantum resources of our protocol. Our measurements reveal that the sensing sensitivity surpasses the standard quantum limit (SQL) with increasing qubit number, reaching 3.78 in a 9-qubit configuration, compared to the SQL of 3.0. The scheme further exhibits inherent robustness to coherent control errors and probed signal noise. This work demonstrates a readily scalable path toward practical quantum sensing advantages with prevalent experimental platforms.

quant-ph↗

arXiv2025

Search for Light Dark Matter with 259-day data in PandaX-4T

We present a search for light dark matter particles through their interactions with atomic electrons and nucleons, utilizing PandaX-4T data with an effective exposure of 1.04 tonne$\cdot$year for ionization-only data and 1.20 tonne$\cdot$year for paired data. Our analysis focuses on the energy range (efficiency$>$0.01) of approximately 0.33 to 3 keV for nuclear recoils, and from 0.04 to 0.39 keV for electronic recoils. We establish the most stringent constraints on spin-independent dark matter-nucleon interactions within a mass range of 2.5 to 5.0 GeV/$c^2$, spin-dependent neutron-only interactions within 2.0 to 5.3 GeV/$c^2$, and spin-dependent proton-only interactions within 2.0 to 3.8 GeV/$c^2$. Their corresponding limits at 3\,GeV/$c^2$ are $1.1 \times 10^{-43}$, $1.6 \times 10^{-38}$, and $5.6 \times 10^{-37}$\,cm$^2$, respectively. Additionally, our results improve the upper limits on the dark matter-electron scattering cross-section by a factor of 1.5 and 9.3 for heavy and light mediator scenarios respectively within 50 MeV/$c^2$ to 10 GeV/$c^2$, compared with previous best results.

hep-ex↗

arXiv2025

MixKVQ: Query-Aware Mixed-Precision KV Cache Quantization for Long-Context Reasoning

Long Chain-of-Thought (CoT) reasoning has significantly advanced the capabilities of Large Language Models (LLMs), but this progress is accompanied by substantial memory and latency overhead from the extensive Key-Value (KV) cache. Although KV cache quantization is a promising compression technique, existing low-bit quantization methods often exhibit severe performance degradation on complex reasoning tasks. Fixed-precision quantization struggles to handle outlier channels in the key cache, while current mixed-precision strategies fail to accurately identify components requiring high-precision representation. We find that an effective low-bit KV cache quantization strategy must consider two factors: a key channel's intrinsic quantization difficulty and its relevance to the query. Based on this insight, we propose MixKVQ, a novel plug-and-play method that introduces a lightweight, query-aware algorithm to identify and preserve critical key channels that need higher precision, while applying per-token quantization for value cache. Experiments on complex reasoning datasets demonstrate that our approach significantly outperforms existing low-bit methods, achieving performance comparable to a full-precision baseline at a substantially reduced memory footprint.

cs.LG↗

arXiv2025

Detection of Performance Interference Among Network Slices in 5G/6G Systems

Recent studies showed that network slices (NSs), which are logical networks supported by shared physical networks, can experience service interference due to sharing of physical and virtual resources. Thus, from the perspective of providing end-to-end (E2E) service quality assurance in 5G/6G systems, it is crucial to discover possible service interference among existing NSs in a timely manner and isolate the potential issues before they can lead to violations of service quality agreements. We study the problem of (a) detecting service interference among NSs in 5G/6G systems and (b) identifying misbehaving NSs and other affected NSs, only using E2E key performance indicator measurements, and propose new algorithms. Our numerical studies demonstrate that, even when the service interference among NSs is weak to moderate, provided that a reasonable number of measurements are available, the proposed algorithms can correctly identify most of shared resources that can lead to service interference among the NSs that utilize the shared resources and misbehaving NSs that can cause potentially adverse service interference and affected NSs.

eess.SP↗

arXiv2025

MMhops-R1: Multimodal Multi-hop Reasoning

The ability to perform multi-modal multi-hop reasoning by iteratively integrating information across various modalities and external knowledge is critical for addressing complex real-world challenges. However, existing Multi-modal Large Language Models (MLLMs) are predominantly limited to single-step reasoning, as existing benchmarks lack the complexity needed to evaluate and drive multi-hop abilities. To bridge this gap, we introduce MMhops, a novel, large-scale benchmark designed to systematically evaluate and foster multi-modal multi-hop reasoning. MMhops dataset comprises two challenging task formats, Bridging and Comparison, which necessitate that models dynamically construct complex reasoning chains by integrating external knowledge. To tackle the challenges posed by MMhops, we propose MMhops-R1, a novel multi-modal Retrieval-Augmented Generation (mRAG) framework for dynamic reasoning. Our framework utilizes reinforcement learning to optimize the model for autonomously planning reasoning paths, formulating targeted queries, and synthesizing multi-level information. Comprehensive experiments demonstrate that MMhops-R1 significantly outperforms strong baselines on MMhops, highlighting that dynamic planning and multi-modal knowledge integration are crucial for complex reasoning. Moreover, MMhops-R1 demonstrates strong generalization to tasks requiring fixed-hop reasoning, underscoring the robustness of our dynamic planning approach. In conclusion, our work contributes a challenging new benchmark and a powerful baseline model, and we will release the associated code, data, and weights to catalyze future research in this critical area.

cs.CV↗

arXiv2025

HHFT: Hierarchical Heterogeneous Feature Transformer for Recommendation Systems

We propose HHFT (Hierarchical Heterogeneous Feature Transformer), a Transformer-based architecture tailored for industrial CTR prediction. HHFT addresses the limitations of DNN through three key designs: (1) Semantic Feature Partitioning: Grouping heterogeneous features (e.g. user profile, item information, behaviour sequennce) into semantically coherent blocks to preserve domain-specific information; (2) Heterogeneous Transformer Encoder: Adopting block-specific QKV projections and FFNs to avoid semantic confusion between distinct feature types; (3) Hiformer Layer: Capturing high-order interactions across features. Our findings reveal that Transformers significantly outperform DNN baselines, achieving a +0.4% improvement in CTR AUC at scale. We have successfully deployed the model on Taobao's production platform, observing a significant uplift in key business metrics, including a +0.6% increase in Gross Merchandise Value (GMV).

cs.IR↗

arXiv2025

Non-abelian extensions and Wells exact sequences of Bol algebras

The aim of this paper is to explore non-abelian extensions of Bol algebras and to study the extensibility of a pair of automorphisms within these non-abelian extensions. We begin by researching non-abelian extensions of Bol algebras and categorizing them based on non-abelian cohomology groups. Additionally, we discuss the conditions under which a pair of automorphisms can be extended in the context of non-abelian extensions of Bol algebras and derive the corresponding Wells exact sequences.

math.RA↗

arXiv2025

IF-Bench: Benchmarking and Enhancing MLLMs for Infrared Images with Generative Visual Prompting

Recent advances in multimodal large language models (MLLMs) have led to impressive progress across various benchmarks. However, their capability in understanding infrared images remains unexplored. To address this gap, we introduce IF-Bench, the first high-quality benchmark designed for evaluating multimodal understanding of infrared images. IF-Bench consists of 499 images sourced from 23 infrared datasets and 680 carefully curated visual question-answer pairs, covering 10 essential dimensions of image understanding. Based on this benchmark, we systematically evaluate over 40 open-source and closed-source MLLMs, employing cyclic evaluation, bilingual assessment, and hybrid judgment strategies to enhance the reliability of the results. Our analysis reveals how model scale, architecture, and inference paradigms affect infrared image comprehension, providing valuable insights for this area. Furthermore, we propose a training-free generative visual prompting (GenViP) method, which leverages advanced image editing models to translate infrared images into semantically and spatially aligned RGB counterparts, thereby mitigating domain distribution shifts. Extensive experiments demonstrate that our method consistently yields significant performance improvements across a wide range of MLLMs. The benchmark and code are available at https://github.com/casiatao/IF-Bench.

cs.CV↗

arXiv2025

Direct Generation of an Array with 78400 Optical Tweezers Using a Single Metasurface

Scalability remains a major challenge in building practical fault-tolerant quantum computers. Currently, the largest number of qubits achieved across leading quantum platforms ranges from hundreds to thousands. In atom arrays, scalability is primarily constrained by the capacity to generate large numbers of optical tweezers, and conventional techniques using acousto-optic deflectors or spatial light modulators struggle to produce arrays much beyond $\sim 10,000$ tweezers. Moreover, these methods require additional microscope objectives to focus the light into micrometer-sized spots, which further complicates system integration and scalability. Here, we demonstrate the experimental generation of an optical tweezer array containing $280\times 280$ spots using a metasurface, nearly an order of magnitude more than most existing systems. The metasurface leverages a large number of subwavelength phase-control pixels to engineer the wavefront of the incident light, enabling both large-scale tweezer generation and direct focusing into micron-scale spots without the need for a microscope. This result shifts the scalability bottleneck for atom arrays from the tweezer generation hardware to the available laser power. Furthermore, the array shows excellent intensity uniformity exceeding $90\%$, making it suitable for homogeneous single-atom loading and paving the way for trapping arrays of more than $10,000$ atoms in the near future.

cond-mat.quant-gas↗

arXiv2025

Nanbeige4-3B Technical Report: Exploring the Frontier of Small Language Models

We present Nanbeige4-3B, a family of small-scale but high-performing language models. Pretrained on 23T high-quality tokens and finetuned on over 30 million diverse instructions, we extend the boundary of the scaling law for small language models. In pre-training, we design a Fine-Grained Warmup-Stable-Decay (FG-WSD) training scheduler, which progressively refines data mixtures across stages to boost model performance. In post-training, to improve the quality of the SFT data, we design a joint mechanism that integrates deliberative generation refinement and chain-of-thought reconstruction, yielding substantial gains on complex tasks. Following SFT, we employ our flagship reasoning model to distill Nanbeige4-3B through our proposed Dual Preference Distillation (DPD) method, which leads to further performance gains. Finally, a multi-stage reinforcement learning phase was applied, leveraging verifiable rewards and preference modeling to strengthen abilities on both reasoning and human alignment. Extensive evaluations show that Nanbeige4-3B not only significantly outperforms models of comparable parameter scale but also rivals much larger models across a wide range of benchmarks. The model checkpoints are available at https://huggingface.co/Nanbeige.

cs.CL↗

arXiv2025

ToolMind Technical Report: A Large-Scale, Reasoning-Enhanced Tool-Use Dataset

Large Language Model (LLM) agents have developed rapidly in recent years to solve complex real-world problems using external tools. However, the scarcity of high-quality trajectories still hinders the development of stronger LLM agents. Most existing works on multi-turn dialogue synthesis validate correctness only at the trajectory level, which may overlook turn-level errors that can propagate during training and degrade model performance. To address these limitations, we introduce ToolMind, a large-scale, high-quality tool-agentic dataset with 160k synthetic data instances generated using over 20k tools and 200k augmented open-source data instances. Our data synthesis pipeline first constructs a function graph based on parameter correlations and then uses a multi-agent framework to simulate realistic user-assistant-tool interactions. Beyond trajectory-level validation, we employ fine-grained turn-level filtering to remove erroneous or suboptimal steps, ensuring that only high-quality reasoning traces are retained. This approach mitigates error amplification during training while preserving self-corrective reasoning signals essential for robust tool-use learning. Models fine-tuned on ToolMind show significant improvements over baselines on several benchmarks.

cs.AI↗

arXiv2025

Visual Reasoning Tracer: Object-Level Grounded Reasoning Benchmark

Recent advances in Multimodal Large Language Models (MLLMs) have significantly improved performance on tasks such as visual grounding and visual question answering. However, the reasoning processes of these models remain largely opaque; they typically output only final predictions without revealing the intermediate steps or fine-grained evidence (e.g., pixels, locations) that lead to the result. This contrasts with human intelligence, which naturally operates through a chain of visual reasoning. To address this limitation, we introduce the Visual Reasoning Tracer (VRT) task, which requires models to not only localize the target object but also explicitly predict the intermediate objects that form the reasoning path. To advance research in this area, we contribute: (1) VRT-Bench, a human-annotated benchmark for evaluating visual reasoning; (2) a new metric for assessing the quality of reasoning traces; and (3) VRT-80k, a large-scale dataset for reasoning model training. Our experiments reveal that while existing models often produce the correct final output, they struggle to ground their intermediate reasoning. In contrast, models trained on VRT-80k achieve substantial improvements in tracing the reasoning path.

cs.CV↗