arXiv subjects

Meng Li

Publications and source records attributed to Meng Li.

At least 109 records · Page 6Linked to original sources

arXiv2026

KEEP: A KV-Cache-Centric Memory Management System for Efficient Embodied Planning

Memory-augmented Large Language Models (LLMs) have demonstrated remarkable capability for complex and long-horizon embodied planning. By keeping track of past experiences and environmental states, memory enables LLMs to maintain a global view, thereby avoiding repetitive exploration. However, existing approaches often store the memory as raw text, leading to excessively long prompts and high prefill latency. While it is possible to store and reuse the KV caches, the efficiency benefits are greatly undermined due to frequent KV cache updates. In this paper, we propose KEEP, a KV-cache-centric memory management system for efficient embodied planning. KEEP features 3 key innovations: (1) a Static-Dynamic Memory Construction algorithm that reduces KV cache recomputation by mixed-granularity memory group; (2) a Multi-hop Memory Re-computation algorithm that dynamically identifies important cross-attention among different memory groups and reconstructs memory interactions iteratively; (3) a Layer-balanced Memory Loading that eliminates unbalanced KV cache loading and cross-attention computation across different layers. Extensive experimental results have demonstrated that KEEP achieves 2.68x speedup with negligible accuracy loss compared with text-based memory methods on ALFRED dataset. Compared with the KV re-computation method CacheBlend (EuroSys'25), KEEP shows 4.13% success rate improvement and 1.90x time-to-first-token (TTFT) reduction. Our code is available on https://github.com/PKU-SEC-Lab/KEEP_Embodied_Memory.

cs.RO↗

arXiv2026

Efficient Document Parsing via Parallel Token Prediction

Document parsing, as a fundamental yet crucial vision task, is being revolutionized by vision-language models (VLMs). However, the autoregressive (AR) decoding inherent to VLMs creates a significant bottleneck, severely limiting parsing speed. In this paper, we propose Parallel-Token Prediction (PTP), a plugable, model-agnostic and simple-yet-effective method that enables VLMs to generate multiple future tokens in parallel with improved sample efficiency. Specifically, we insert some learnable tokens into the input sequence and design corresponding training objectives to equip the model with parallel decoding capabilities for document parsing. Furthermore, to support effective training, we develop a comprehensive data generation pipeline that efficiently produces large-scale, high-quality document parsing training data for VLMs. Extensive experiments on OmniDocBench and olmOCR-bench demonstrate that our method not only significantly improves decoding speed (1.6x-2.2x) but also reduces model hallucinations and exhibits strong generalization abilities.

cs.CL↗

arXiv2026

Towards unified brain-to-text decoding across speech production and perception

Speech production and perception are the main ways humans communicate daily. Prior brain-to-text decoding studies have largely focused on a single modality and alphabetic languages. Here, we present a unified brain-to-sentence decoding framework for both speech production and perception in Mandarin Chinese. The framework exhibits strong generalization ability, enabling sentence-level decoding when trained only on single-character data and supporting characters and syllables unseen during training. In addition, it allows direct and controlled comparison of neural dynamics across modalities. Mandarin speech is decoded by first classifying syllable components in Hanyu Pinyin, namely initials and finals, from neural signals, followed by a post-trained large language model (LLM) that maps sequences of toneless Pinyin syllables to Chinese sentences. To enhance LLM decoding, we designed a three-stage post-training and two-stage inference framework based on a 7-billion-parameter LLM, achieving overall performance that exceeds larger commercial LLMs with hundreds of billions of parameters or more. In addition, several characteristics were observed in Mandarin speech production and perception: speech production involved neural responses across broader cortical regions than auditory perception; channels responsive to both modalities exhibited similar activity patterns, with speech perception showing a temporal delay relative to production; and decoding performance was broadly comparable across hemispheres. Our work not only establishes the feasibility of a unified decoding framework but also provides insights into the neural characteristics of Mandarin speech production and perception. These advances contribute to brain-to-text decoding in logosyllabic languages and pave the way toward neural language decoding systems supporting multiple modalities.

q-bio.NC↗

arXiv2026

Development of a dual-phase xenon time projection chamber prototype for the RELICS experiment

The RELICS (REactor neutrino LIquid xenon Coherent elastic Scattering) experiment aims to detect coherent elastic neutrino-nucleus scattering from reactor antineutrinos using a dual-phase xenon time projection chamber. To validate the detector concept and ensure technical reliability for the full-scale experiment, a dedicated prototype was designed, constructed, and operated. This work presents an overview of the design, construction, and operational performance of the prototype, with emphasis on its major subsystems, including the TPC, cryogenic and xenon purification systems, slow control, and data acquisition. During operation, the detector demonstrated the capability to achieve a sub-keV energy threshold required for the RELICS physics program, as reflected by a measured single electron gain of 34.30~$\pm$~0.01~(stat.)~PE/e$^-$ and the successful detection of 0.27~keV L-shell decay events from $^{37}$Ar. In addition, essential data analysis techniques and simulation frameworks were developed and validated, establishing the methodological foundation for future RELICS operations. The successful construction and operation of this prototype confirm the feasibility of the core technologies and provide a crucial experimental basis for the final RELICS detector.

physics.ins-det↗

arXiv2026

DualSpec: Accelerating Deep Research Agents via Dual-Process Action Speculation

Large language model-based deep research agents have been increasingly popular for addressing long-horizon information-seeking tasks, but they often incur high end-to-end latency due to extensive reasoning and frequent tool use. Speculation frameworks aim to reduce latency by overlapping action execution with reasoning; however, existing approaches typically rely on uniform speculation strategies and strict action matching, which limits inference speedups and robustness. In this work, we revisit the speculate-verify paradigm for deep research agents through the lens of action heterogeneity. We show that \textit{Search} and \textit{Visit} actions exhibit fundamentally different reasoning and model capacity requirements: entropy-based analysis reveals that Search decisions have higher uncertainty and benefit significantly from explicit reasoning, whereas Visit decisions have lower entropy and depend primarily on model capacity. Motivated by this dual-process characteristic, we propose DualSpec, a heterogeneous speculation framework equipped with a lightweight, confidence-based semantic verifier. Experiments across multiple models and benchmarks demonstrate that DualSpec achieves up to 3.28$\times$ end-to-end speedup while maintaining accuracy comparable to fully reasoning agents.

cs.LG↗

arXiv2026

Pre/Absence: Prompting Cultural Awareness and Understanding for Lost Architectural Heritage in Virtual Reality

Lost architectural heritage presents interpretive challenges due to vanished structures and fragmented historical records. Using Hanyuan Hall of the Tang dynasty's Daming Palace as a case study, we conducted a formative investigation with archaeologists, heritage administrators, and visitors to identify key issues in current interpretation practices. We found that these practices often compress complex cultural layers into factual summaries and rely on linear narratives that overlook the continuing reinterpretations following a site's disappearance. In response, we designed Pre/Absence, a virtual reality experience grounded in the presence-absence dialectic to interweave tangible and vanished aspects of heritage within a spatiotemporal narrative. A mixed-method study with 28 participants compared Pre/Absence to a paper-based experience. Both improved users' factual understanding, but the VR experience more strongly enhanced cultural awareness, evoked emotional engagement with loss, and encouraged critical reflection on the evolving social and political meanings of heritage. The findings suggest that VR can move beyond static reconstruction to engage users as co-constructors of cultural meaning, providing a nuanced framework for critical heritage narrative design in human-computer interaction.

cs.HC↗

arXiv2026

Estimation and Hypothesis Testing of Derivatives in Smoothing Spline ANOVA Models

Within the framework of smoothing spline ANOVA, we propose a plug-in kernel ridge regression estimator to estimate the derivatives of the underlying multivariate regression function. We first establish an $L_\infty$ convergence rate of the proposed estimator under general random designs. When the covariates are uniformly distributed, we provide a in-depth analysis that includes a sharp upper bound and the minimax lower bound of the $L_2$ convergence rate. Additionally, motivated by a wide range of applications, we propose a hypothesis testing procedure to examine whether a derivative is zero. Theoretical results demonstrate that the proposed testing procedure achieves the correct size under the null hypothesis and is asymptotically powerful under local alternatives. For ease of use, we also develop an associated bootstrap algorithm to construct the rejection region and calculate p-value, and the consistency of the proposed algorithm is established. Simulation studies using synthetic data and an application to a real-world dataset confirm the effectiveness of our approach.

stat.ME↗

arXiv2026

RoboMIND 2.0: A Multimodal, Bimanual Mobile Manipulation Dataset for Generalizable Embodied Intelligence

While data-driven imitation learning has revolutionized robotic manipulation, current approaches remain constrained by the scarcity of large-scale, diverse real-world demonstrations. Consequently, the ability of existing models to generalize across long-horizon bimanual tasks and mobile manipulation in unstructured environments remains limited. To bridge this gap, we present RoboMIND 2.0, a comprehensive real-world dataset comprising over 310K dual-arm manipulation trajectories collected across six distinct robot embodiments and 739 complex tasks. Crucially, to support research in contact-rich and spatially extended tasks, the dataset incorporates 12K tactile-enhanced episodes and 20K mobile manipulation trajectories. Complementing this physical data, we construct high-fidelity digital twins of our real-world environments, releasing an additional 20K-trajectory simulated dataset to facilitate robust sim-to-real transfer. To fully exploit the potential of RoboMIND 2.0, we propose MIND-2 system, a hierarchical dual-system frame-work optimized via offline reinforcement learning. MIND-2 integrates a high-level semantic planner (MIND-2-VLM) to decompose abstract natural language instructions into grounded subgoals, coupled with a low-level Vision-Language-Action executor (MIND-2-VLA), which generates precise, proprioception-aware motor actions.

cs.RO↗

arXiv2026

Direct Spectral Acceleration of First-Order Methods for Saddle Point Problems with Bilinear Coupling

We study convex-concave saddle point problems with bilinear coupling, covering linearly constrained convex optimization and more general nonsmooth or constrained models via a proximable term in the dual objective. In linearly convergent regimes, we characterize how spectral properties of the coupling matrix and objective conditioning jointly determine the attainable linear rates. We propose direct spectral acceleration for first-order primal--dual methods for a class of bilinear-coupled saddle point problems, including affinely constrained smooth strongly convex optimization and extensions with proximable dual terms. The resulting algorithms distinguish objective-dominated and coupling matrix-dominated regimes and attain optimal linear convergence without Chebyshev inner loops or double-loop designs. We further develop stochastic block-coordinate extensions in the affinely constrained case with separable objectives; we also establish optimal linear rates matching the block-coordinate lower bound. For both deterministic and stochastic methods, we provide matching worst-case lower bounds via explicit finite-dimensional hard instances.

math.OC↗

arXiv2026

UFO: Unlocking Ultra-Efficient Quantized Private Inference with Protocol and Algorithm Co-Optimization

Private convolutional neural network (CNN) inference based on secure two-party computation (2PC) suffers from high communication and latency overhead, especially from convolution layers. In this paper, we propose UFO, a quantized 2PC inference framework that jointly optimizes the 2PC protocols and quantization algorithm. UFO features a novel 2PC protocol that systematically combines the efficient Winograd convolution algorithm with quantization to improve inference efficiency. However, we observe that naively combining quantization and Winograd convolution faces the following challenges: 1) From the inference perspective, Winograd transformations introduce extensive additions and require frequent bit width conversions to avoid inference overflow, leading to non-negligible communication overhead; 2) From the training perspective, Winograd transformations introduce weight outliers that make quantization-aware training (QAT) difficult, resulting in inferior model accuracy. To address these challenges, we co-optimize both protocol and algorithm. 1) At the protocol level, we propose a series of graph-level optimizations for 2PC inference to minimize the communication. 2) At the algorithm level, we develop a mixed-precision QAT algorithm based on layer sensitivity to optimize model accuracy given communication constraints. To accommodate the outliers, we further introduce a 2PC-friendly bit re-weighting algorithm to increase the representation range without explicitly increasing bit widths. With extensive experiments, UFO demonstrates 11.7x, 3.6x, and 6.3x communication reduction with 1.29%, 1.16%, and 1.29% higher accuracy compared to state-of-the-art frameworks SiRNN, COINN, and CoPriv, respectively.

cs.CR↗

arXiv2026

AXLearn: Modular, Hardware-Agnostic Large Model Training

AXLearn is a production system which facilitates scalable and high-performance training of large deep learning models. Compared to other state-of-art deep learning systems, AXLearn has a unique focus on modularity and support for hardware-agnostic training. AXLearn's internal interfaces between software components follow strict encapsulation, allowing different components to be assembled to facilitate rapid model development and experimentation on different hardware infrastructure. AXLearn maintains constant complexity as we scale the components in the system, compared to linear or quadratic complexity in state-of-the-art training systems. This allows integrating features such as Rotary Position Embeddings (RoPE) into AXLearn across hundred of modules with just 10 lines of code, compared to hundreds as required in other systems. At the same time, AXLearn maintains equivalent performance compared to state-of-the-art training systems. Finally, we share our experience in the development and operation of AXLearn at Apple.

cs.LG↗

arXiv2026

Design and characterization of a photosensor system for the RELICS experiment

In this paper, we present the design and characterization of a photosensor system developed for the RELICS experiment. An extended dynamic range base was designed to mitigate photomultiplier tube (PMT) saturation caused by intense cosmic muon backgrounds in the surface-level RELICS detector. The system employs dual readout from the anode and the seventh dynode to extend the linear response range of the PMT. In particular, our characterization and measurements of Hamamatsu R8520-406 PMTs confirm stable operation under positive high-voltage bias, extending the linear response range by more than an order of magnitude. Furthermore, a model of PMT saturation and recovery was developed to evaluate the influence of cosmic muon signals in the RELICS detector. The results demonstrate the system capability to detect coherent elastic neutrino-nucleus scattering signals under surface-level cosmic backgrounds, and suggest the potential to extend the scientific reach of RELICS to MeV-scale interactions.

physics.ins-det↗

arXiv2026

FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency

Generative modeling-based visuomotor policies have been widely adopted in robotic manipulation, attributed to their ability to model multimodal action distributions. However, the high inference cost of multi-step sampling limits its applicability in real-time robotic systems. Existing approaches accelerate sampling in generative modeling-based visuomotor policies by adapting techniques originally developed to speed up image generation. However, a major distinction exists: image generation typically produces independent samples without temporal dependencies, while robotic manipulation requires generating action trajectories with continuity and temporal coherence. To this end, we propose FreqPolicy, a novel approach that first imposes frequency consistency constraints on flow-based visuomotor policies. Our work enables the action model to capture temporal structure effectively while supporting efficient, high-quality one-step action generation. Concretely, we introduce a frequency consistency constraint objective that enforces alignment of frequency-domain action features across different timesteps along the flow, thereby promoting convergence of one-step action generation toward the target distribution. In addition, we design an adaptive consistency loss to capture structural temporal variations inherent in robotic manipulation tasks. We assess FreqPolicy on 53 tasks across 3 simulation benchmarks, proving its superiority over existing one-step action generators. We further integrate FreqPolicy into the vision-language-action (VLA) model and achieve acceleration without performance degradation on 40 tasks of LIBERO. Besides, we show efficiency and effectiveness in real-world robotic scenarios with an inference frequency of 93.5 Hz.

cs.RO↗

arXiv2026

Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache

Generative large language models (LLMs) have revolutionized multiple domains. Modern LLMs predominantly rely on an autoregressive decoding strategy, which generates output tokens sequentially and employs a key-value cache (KV cache) to avoid redundant computation. However, the widespread deployment of LLMs has raised serious privacy concerns, as users are feeding all types of data into the model, motivating the development of secure inference frameworks based on fully homomorphic encryption (FHE). A major limitation of existing FHE-based frameworks is their inability to effectively integrate the KV cache, resulting in prohibitively high latency for autoregressive decoding. In this paper, we propose Cachemir, a KV Cache Accelerated Homomorphic Encrypted LLM Inference Regime to overcome this limitation. Cachemir comprises three key technical contributions: 1) a set of novel HE packing algorithms specifically designed to leverage the computational advantages of the KV cache; 2) an interleaved replicated packing algorithm to efficiently compute the vector-matrix multiplications that result from using the KV cache in Transformer linear layers; and 3) an augmented bootstrapping placement strategy that accounts for the KV cache to minimize bootstrapping cost. We demonstrate that Cachemir achieves $48.83\times$ and $67.16\times$ speedup over MOAI (ICML'25) and THOR (CCS'25) respectively on CPU and consumes less than 100 seconds on GPU to generate an output token for Llama-3-8B.

cs.CR↗

arXiv2026

Adaptive Attribute-Decoupled Encryption for Trusted Respiratory Monitoring in Resource-Limited Consumer Healthcare

Respiratory monitoring is an extremely important task in modern medical services. Due to its significant advantages, e.g., non-contact, radar-based respiratory monitoring has attracted widespread attention from both academia and industry. Unfortunately, though it can achieve high monitoring accuracy, consumer electronics-grade radar data inevitably contains User-sensitive Identity Information (USI), which may be maliciously used and further lead to privacy leakage. To track these challenges, by variational mode decomposition (VMD) and adversarial loss-based encryption, we propose a novel Trusted Respiratory Monitoring paradigm, Tru-RM, to perform automated respiratory monitoring through radio signals while effectively anonymizing USI. The key enablers of Tru-RM are Attribute Feature Decoupling (AFD), Flexible Perturbation Encryptor (FPE), and robust Perturbation Tolerable Network (PTN) used for attribute decomposition, identity encryption, and perturbed respiratory monitoring, respectively. Specifically, AFD is designed to decompose the raw radar signals into the universal respiratory component, the personal difference component, and other unrelated components. Then, by using large noise to drown out the other unrelated components, and the phase noise algorithm with a learning intensity parameter to eliminate USI in the personal difference component, FPE is designed to achieve complete user identity information encryption without affecting respiratory features. Finally, by designing the transferred generalized domain-independent network, PTN is employed to accurately detect respiration when waveforms change significantly. Extensive experiments based on various detection distances, respiratory patterns, and durations demonstrate the superior performance of Tru-RM on strong anonymity of USI, and high detection accuracy of perturbed respiratory waveforms.

cs.CR↗

arXiv2026

Full end-to-end diagnostic workflow automation of 3D OCT via foundation model-driven AI for retinal diseases

Optical coherence tomography (OCT) has revolutionized retinal disease diagnosis with its high-resolution and three-dimensional imaging nature, yet its full diagnostic automation in clinical practices remains constrained by multi-stage workflows and conventional single-slice single-task AI models. We present Full-process OCT-based Clinical Utility System (FOCUS), a foundation model-driven framework enabling end-to-end automation of 3D OCT retinal disease diagnosis. FOCUS sequentially performs image quality assessment with EfficientNetV2-S, followed by abnormality detection and multi-disease classification using a fine-tuned Vision Foundation Model. Crucially, FOCUS leverages a unified adaptive aggregation method to intelligently integrate 2D slices-level predictions into comprehensive 3D patient-level diagnosis. Trained and tested on 3,300 patients (40,672 slices), and externally validated on 1,345 patients (18,498 slices) across four different-tier centers and diverse OCT devices, FOCUS achieved high F1 scores for quality assessment (99.01%), abnormally detection (97.46%), and patient-level diagnosis (94.39%). Real-world validation across centers also showed stable performance (F1: 90.22%-95.24%). In human-machine comparisons, FOCUS matched expert performance in abnormality detection (F1: 95.47% vs 90.91%) and multi-disease diagnosis (F1: 93.49% vs 91.35%), while demonstrating better efficiency. FOCUS automates the image-to-diagnosis pipeline, representing a critical advance towards unmanned ophthalmology with a validated blueprint for autonomous screening to enhance population scale retinal care accessibility and efficiency.

cs.CV↗

arXiv2026

IntraSlice: Towards High-Performance Structural Pruning with Block-Intra PCA for LLMs

Large Language Models (LLMs) achieve strong performance across diverse tasks but face deployment challenges due to their massive size. Structured pruning offers acceleration benefits but leads to significant performance degradation. Recent PCA-based pruning methods have alleviated this issue by retaining key activation components, but are only applied between modules in order to fuse the transformation matrix, which introduces extra parameters and severely disrupts activation distributions due to residual connections. To address these issues, we propose IntraSlice, a framework that applies block-wise module-intra PCA compression pruning. By leveraging the structural characteristics of Transformer modules, we design an approximate PCA method whose transformation matrices can be fully fused into the model without additional parameters. We also introduce a PCA-based global pruning ratio estimator that further considers the distribution of compressed activations, building on conventional module importance. We validate our method on Llama2, Llama3, and Phi series across various language benchmarks. Experimental results demonstrate that our approach achieves superior compression performance compared to recent baselines at the same compression ratio or inference speed.

cs.LG↗

arXiv2026

LMTE: Putting the "Reasoning" into WAN Traffic Engineering with Language Models

The rapid expansion of modern wide-area networks (WANs) has made traffic engineering (TE) increasingly challenging, as traditional solvers struggle to keep pace. Although existing offline ML-driven approaches accelerate TE optimization with deep neural networks (DNNs), they often lack sufficient expressiveness and generalization on unseen traffic patterns or topologies, limiting their practicality. Inspired by the success of large language models (LMs), for the first time, this paper investigates their potential as general-purpose traffic planners. Our contributions are two-fold: (i) Theoretically, we show that pre-trained LMs can simulate the sequential decision processes underlying TE and, crucially, exhibit parallel reasoning capabilities, making them well-suited for the task; (ii) Practically, we present LMTE, a novel LM-driven TE framework that embraces these insights through efficient multimodal alignment and lightweight configuration generation, all while preserving the model's original abilities. Extensive experiments demonstrate that fold matches top-tier performance on five datasets, achieving up to 15\% better maximum link utilization (MLU) and consistently lower performance degradation across diverse scenarios, e.g., less than 5\% with high traffic dynamics and link failures. Moreover, it achieves 10 to 100 times speedups over traditional TE solvers. To aid future works, our codebase is available at https://github.com/Y-debug-sys/LMTE.

cs.NI↗