arXiv subjects

Huan Li

Publications and source records attributed to Huan Li.

At least 55 records · Page 3Linked to original sources

arXiv2026

RedParrot: Accelerating NL-to-DSL for Business Analytics via Query Semantic Caching

Recently, at Xiaohongshu, the rapid expansion of e-commerce and advertising demands real-time business analytics with high accuracy and low latency. To meet this demand, systems typically rely on converting natural language (NL) queries into Domain-Specific Languages (DSLs) to ensure semantic consistency, validation, and portability. However, existing multi-stage LLM pipelines for this NL-to-DSL task suffer from prohibitive latency, high cost, and error propagation, rendering them unsuitable for enterprise-scale deployment. In this paper, we propose RedParrot, a novel NL-to-DSL framework that accelerates inference via a semantic cache. Observing the high repetition and stable structural patterns in user queries, RedParrot bypasses the costly pipeline by matching new requests against cached "query skeletons" (normalized structural patterns) and adapting their corresponding DSLs. Our core technical contributions include (1) an offline skeleton construction strategy, (2) an online, entity-agnostic embedding model trained via contrastive learning for robust matching, and (3) a heterogeneous Retrieval-Augmented Generation (RAG) method that integrates diverse knowledge sources to handle unseen entities. Experiments on six real enterprise datasets from Xiaohongshu show RedParrot achieves an average 3.6x speedup and an 8.26% accuracy improvement. Furthermore, on new public benchmarks adapted from Spider and BIRD, it boosts accuracy by 34.8%, substantially outperforming standard in-context learning baselines.

cs.IR↗

arXiv2026

DQE: A Semantic-Aware Evaluation Metric for Time Series Anomaly Detection

Time series anomaly detection has achieved remarkable progress in recent years. However, evaluation practices have received comparatively less attention, despite their critical importance. Existing metrics exhibit several limitations: (1) bias toward point-level coverage, (2) insensitivity or inconsistency in near-miss detections, (3) inadequate penalization of false alarms, and (4) inconsistency caused by threshold or threshold-interval selection. These limitations can produce unreliable or counterintuitive results, hindering objective progress. In this work, we revisit the evaluation of time series anomaly detection from the perspective of detection semantics and propose a novel metric for more comprehensive assessment. We first introduce a partitioning strategy grounded in detection semantics, which decomposes the local temporal region of each anomaly into three functionally distinct subregions. Using this partitioning, we evaluate overall detection behavior across events and design finer-grained scoring mechanisms for each subregion, enabling more reliable and interpretable assessment. Through a systematic study of existing metrics, we identify an evaluation bias associated with threshold-interval selection and adopt an approach that aggregates detection qualities across the full threshold spectrum, thereby eliminating evaluation inconsistency. Extensive experiments on synthetic and real-world data demonstrate that our metric provides stable, discriminative, and interpretable evaluation, while achieving robust assessment compared with ten widely used metrics.

cs.LG↗

arXiv2026

Novel High-Scalability Architecture for Photonic Deep Learning

Photonic computing promises ultrafast and energy-efficient artificial intelligence. However, existing photonic neural networks (PNNs) remain functionally shallow and difficult to scale. Here we establish a theory-guided framework showing that power stability and complex-field correlation are the fundamental prerequisites for scalable, coherent PNNs. Building on these macroscopic principles, we introduce the Coherent, Compensated and Cross-connected (C3) unit - an architecture that integrates coherent nonlinearity, active loss compensation and native optical residual connectivity. Implemented on a silicon-on-insulator platform, the C3 unit provides reconfigurable activation functions and dynamic energy stabilization without external amplification. We validate this framework using a width-constrained spiral benchmark, in which the C3 unit substantially improves parameter utilization and power robustness relative to incoherent nonlinearities. In a high-complexity 1,623-class recognition task, our C3-enabled coherent residual network (CoP-ResNet) achieves a top-1 accuracy of 77.92%, whereas non-residual architectures fail to converge. Together, these results offer a physically grounded, theory-guided pathway toward greater optical processing depth, laying the foundation for next-generation, large-scale photonic computing architectures.

physics.optics↗

arXiv2026

VERA: Identifying and Leveraging Visual Evidence Retrieval Heads in Long-Context Understanding

While Vision-Language Models (VLMs) have shown promise in textual understanding, they face significant challenges when handling long context and complex reasoning tasks. In this paper, we dissect the internal mechanisms governing long-context processing in VLMs to understand their performance bottlenecks. Through the lens of attention analysis, we identify specific Visual Evidence Retrieval (VER) Heads - a sparse, dynamic set of attention heads critical for locating visual cues during reasoning, distinct from static OCR heads. We demonstrate that these heads are causal to model performance; masking them leads to significant degradation. Leveraging this discovery, we propose VERA (Visual Evidence Retrieval Augmentation), a training-free framework that detects model uncertainty (i.e., entropy) to trigger the explicit verbalization of visual evidence attended by VER heads. Comprehensive experiments demonstrate that VERA significantly improves long-context understanding of open-source VLMs: it yields an average relative improvement of 21.3% on Qwen3-VL-8B-Instruct and 20.1% on GLM-4.1V-Thinking across five benchmarks.

cs.CV↗

arXiv2026

SafeLoad: Efficient Admission Control Framework for Identifying Memory-Overloading Queries in Cloud Data Warehouses

Memory overload is a common form of resource exhaustion in cloud data warehouses. When database queries fail due to memory overload, it not only wastes critical resources such as CPU time but also disrupts the execution of core business processes, as memory-overloading (MO) queries are typically part of complex workflows. If such queries are identified in advance and scheduled to memory-rich serverless clusters, it can prevent resource wastage and query execution failure. Therefore, cloud data warehouses desire an admission control framework with high prediction precision, interpretability, efficiency, and adaptability to effectively identify MO queries. However, existing admission control frameworks primarily focus on scenarios like SLA satisfaction and resource isolation, with limited precision in identifying MO queries. Moreover, there is a lack of publicly available MO-labeled datasets with workloads for training and benchmarking. To tackle these challenges, we propose SafeLoad, the first query admission control framework specifically designed to identify MO queries. Alongside, we release SafeBench, an open-source, industrial-scale benchmark for this task, which includes 150 million real queries. SafeLoad first filters out memory-safe queries using the interpretable discriminative rule. It then applies a hybrid architecture that integrates both a global model and cluster-level models, supplemented by a misprediction correction module to identify MO queries. Additionally, a self-tuning quota management mechanism dynamically adjusts prediction quotas per cluster to improve precision. Experimental results show that SafeLoad achieves state-of-the-art prediction performance with low online and offline time overhead. Specifically, SafeLoad improves precision by up to 66% over the best baseline and reduces wasted CPU time by up to 8.09x compared to scenarios without SafeLoad.

cs.DB↗

arXiv2026

DeXOR: Enabling XOR in Decimal Space for Streaming Lossless Compression of Floating-point Data

With streaming floating-point numbers being increasingly prevalent, effective and efficient compression of such data is critical. Compression schemes must be able to exploit the similarity, or smoothness, of consecutive numbers and must be able to contend with extreme conditions, such as high-precision values or the absence of smoothness. We present DeXOR, a novel framework that enables decimal XOR procedure to encode decimal-space longest common prefixes and suffixes, achieving optimal prefix reuse and effective redundancy elimination. To ensure accurate and low-cost decompression even with binary-decimal conversion errors, DeXOR incorporates 1) scaled truncation with error-tolerant rounding and 2) different bit management strategies optimized for decimal XOR. Additionally, a robust exception handler enhances stability by managing floating-point exponents, maintaining high compression ratios under extreme conditions. In evaluations across 22 datasets, DeXOR surpasses state-of-the-art schemes, achieving a 15% higher compression ratio and a 20% faster decompression speed while maintaining a competitive compression speed. DeXOR also offers scalability under varying conditions and exhibits robustness in extreme scenarios where other schemes fail.

cs.DB↗

arXiv2025

Analyzing the Mechanism of Attention Collapse in VGGT from a Dynamics Perspective

Visual Geometry Grounded Transformer (VGGT) delivers state-of-the-art feed-forward 3D reconstruction, yet its global self-attention layer suffers from a drastic collapse phenomenon when the input sequence exceeds a few hundred frames: attention matrices rapidly become near rank-one, token geometry degenerates to an almost one-dimensional subspace, and reconstruction error accumulates super-linearly.In this report,we establish a rigorous mathematical explanation of the collapse by viewing the global-attention iteration as a degenerate diffusion process.We prove that,in VGGT, the token-feature flow converges toward a Dirac-type measure at a $O(1/L)$ rate, where $L$ is the layer index, yielding a closed-form mean-field partial differential equation that precisely predicts the empirically observed rank profile.The theory quantitatively matches the attention-heat-map evolution and a series of experiments outcomes reported in relevant works and explains why its token-merging remedy -- which periodically removes redundant tokens -- slows the effective diffusion coefficient and thereby delays collapse without additional training.We believe the analysis provides a principled lens for interpreting future scalable 3D-vision transformers,and we highlight its potential for multi-modal generalization.

cs.CV↗

arXiv2025

MovSemCL: Movement-Semantics Contrastive Learning for Trajectory Similarity (Extension)

Trajectory similarity computation is fundamental functionality that is used for, e.g., clustering, prediction, and anomaly detection. However, existing learning-based methods exhibit three key limitations: (1) insufficient modeling of trajectory semantics and hierarchy, lacking both movement dynamics extraction and multi-scale structural representation; (2) high computational costs due to point-wise encoding; and (3) use of physically implausible augmentations that distort trajectory semantics. To address these issues, we propose MovSemCL, a movement-semantics contrastive learning framework for trajectory similarity computation. MovSemCL first transforms raw GPS trajectories into movement-semantics features and then segments them into patches. Next, MovSemCL employs intra- and inter-patch attentions to encode local as well as global trajectory patterns, enabling efficient hierarchical representation and reducing computational costs. Moreover, MovSemCL includes a curvature-guided augmentation strategy that preserves informative segments (e.g., turns and intersections) and masks redundant ones, generating physically plausible augmented views. Experiments on real-world datasets show that MovSemCL is capable of outperforming state-of-the-art methods, achieving mean ranks close to the ideal value of 1 at similarity search tasks and improvements by up to 20.3% at heuristic approximation, while reducing inference latency by up to 43.4%.

cs.CV↗

arXiv2025

A Joint Auction Framework with Externalities and Adaptation

Recently, joint advertising has gained significant attention as an effective approach to enhancing the efficiency and revenue of advertising slot allocation. Unlike traditional advertising, which allocates advertising slots exclusively to a single advertiser, joint advertising displays advertisements from brands and stores that have established a joint selling relationship within the same advertising slot. However, existing approaches often struggle to accommodate both joint and traditional advertising frameworks, thereby limiting the revenue potential and generalizability of joint advertising. Furthermore, these methods are constrained by two critical limitations: they generally neglect the influence of global externalities, and they fail to address the bidding variability stemming from multi-party advertiser participation. Collectively, these limitations present substantial challenges to the design of joint auction mechanisms. To address these challenges, we propose a Joint Auction Framework incorporating Externalities and Adaptation, and leverage the automated mechanism design (AMD) method through our proposed JEANet to compute joint auction mechanisms that satisfy the conditions of individual rationality (IR) and approximate dominant strategy incentive compatibility (DSIC). As the first AMD method to integrate global externalities into joint auctions, JEANet dynamically adapts to the bidding characteristics of multi-party advertiser and enables unified auctions that integrate both joint and traditional advertising. Extensive experimental results demonstrate that JEANet outperforms state-of-the-art baselines in multi-slot joint auctions.

cs.GT↗

arXiv2025

CTkvr: KV Cache Retrieval for Long-Context LLMs via Centroid then Token Indexing

Large language models (LLMs) are increasingly applied in long-context scenarios such as multi-turn conversations. However, long contexts pose significant challenges for inference efficiency, including high memory overhead from Key-Value (KV) cache and increased latency due to excessive memory accesses. Recent methods for dynamic KV selection struggle with trade-offs: block-level indexing degrades accuracy by retrieving irrelevant KV entries, while token-level indexing incurs high latency from inefficient retrieval mechanisms. In this paper, we propose CTKVR, a novel centroid-then-token KV retrieval scheme that addresses these limitations. CTKVR leverages a key observation: query vectors adjacent in position exhibit high similarity after Rotary Position Embedding (RoPE) and share most of their top-k KV cache entries. Based on this insight, CTKVR employs a two-stage retrieval strategy: lightweight centroids are precomputed during prefilling for centroid-grained indexing, followed by token-level refinement for precise KV retrieval. This approach balances retrieval efficiency and accuracy. To further enhance performance, we implement an optimized system for indexing construction and search using CPU-GPU co-execution. Experimentally, CTKVR achieves superior performance across multiple benchmarks with less than 1% accuracy degradation. Meanwhile, CTKVR delivers 3 times and 4 times throughput speedups on Llama-3-8B and Yi-9B at 96K context length across diverse GPU hardware.

cs.CL↗

arXiv2025

Not All Data are Good Labels: On the Self-supervised Labeling for Time Series Forecasting

Time Series Forecasting (TSF) is a crucial task in various domains, yet existing TSF models rely heavily on high-quality data and insufficiently exploit all available data. This paper explores a novel self-supervised approach to re-label time series datasets by inherently constructing candidate datasets. During the optimization of a simple reconstruction network, intermediates are used as pseudo labels in a self-supervised paradigm, improving generalization for any predictor. We introduce the Self-Correction with Adaptive Mask (SCAM), which discards overfitted components and selectively replaces them with pseudo labels generated from reconstructions. Additionally, we incorporate Spectral Norm Regularization (SNR) to further suppress overfitting from a loss landscape perspective. Our experiments on eleven real-world datasets demonstrate that SCAM consistently improves the performance of various backbone models. This work offers a new perspective on constructing datasets and enhancing the generalization of TSF models through self-supervised learning. The code is available at https://github.com/SuDIS-ZJU/SCAM.

cs.LG↗

arXiv2025

Switching of topological phase transition from semiconductor to ideal Weyl states in Cu$_2$SnSe$_3$ family of materials

The exploration of topological phase transition (TPT) mechanisms constitutes a central theme in quantum materials research. Conventionally, transitions between Weyl semimetals (WSMs) and other topological states rely on the breaking of time-reversal symmetry (TRS) or precise manipulation of lattice symmetry, thus constraints the available control strategies and restrict the scope of viable material systems. In this work, we propose a novel mechanism for TPT that operates without TRS breaking or lattice symmetry modification: a class of semiconductors can be directly transformed into an ideal WSM via bandgap closure. This transition is driven by chemical doping, which simultaneously modulates the band gap and enhances spin-orbit coupling (SOC), leading to band inversion between the valence and conduction bands and thereby triggering the TPT. Using first-principles calculations, we demonstrate the feasibility of this mechanism in the Cu$_2$SnSe$_3$ family of materials, where two pairs of Weyl points emerge with energies extremely close to the Fermi level. The bulk Fermi surface becomes nearly point-like, while the surface Fermi surface consists exclusively of Fermi arcs. This symmetry-independent mechanism bypasses the constraints of conventional symmetry-based engineering, and also offers an ideal platform for probing the anomalous transport properties of WSMs.

cond-mat.mtrl-sci↗

arXiv2025

Pockels effect induced strong Kerr nonlinearity in a lithium niobate waveguide

The utilization of Kerr nonlinearity in lithium niobate has been extensively investigated over the years. Nevertheless, the practical implementation of Kerr nonlinearity in waveguides has been constrained by the material's inherently low third-order nonlinear coefficients. Here, we present a significant advancement by demonstrating Pockels effect-induced strong Kerr nonlinearity in a periodically poled thin-film lithium niobate waveguide. Both effective four-wave mixing (FWM) and cascaded effective FWM processes are experimentally observed. The induced FWM process achieves a remarkable maximum output power of -8.5 dBm, spanning a wavelength spectrum of over 116.8 nm. Analysis reveals that the induced effective Kerr nonlinearity exhibits a substantial effective nonlinear refractive index as $2.9\times 10^{-15} m^{2}W^{-1}$, corresponding to an effective nonlinear refractive index enhancement factor of $1.6\times 10^{4}$ relative to the intrinsic value. Moreover, a wavelength-converting experiment demonstrates a flat optic-to-optic response over a broadband radiofrequency spectrum, confirming that signal integrity is well preserved after on-chip effective FWM conversion. Therefore, the demonstrated efficient and broadband Pockels effect induced effective Kerr nonlinearity paves the way for novel applications in diverse fields, including spectroscopy, parametric amplification, quantum correlation studies, and wavelength conversion technologies.

physics.optics↗

arXiv2025

AnyTalker: Scaling Multi-Person Talking Video Generation with Interactivity Refinement

Recently, multi-person video generation has started to gain prominence. While a few preliminary works have explored audio-driven multi-person talking video generation, they often face challenges due to the high costs of diverse multi-person data collection and the difficulty of driving multiple identities with coherent interactivity. To address these challenges, we propose AnyTalker, a multi-person generation framework that features an extensible multi-stream processing architecture. Specifically, we extend Diffusion Transformer's attention block with a novel identity-aware attention mechanism that iteratively processes identity-audio pairs, allowing arbitrary scaling of drivable identities. Besides, training multi-person generative models demands massive multi-person data. Our proposed training pipeline depends solely on single-person videos to learn multi-person speaking patterns and refines interactivity with only a few real multi-person clips. Furthermore, we contribute a targeted metric and dataset designed to evaluate the naturalness and interactivity of the generated multi-person videos. Extensive experiments demonstrate that AnyTalker achieves remarkable lip synchronization, visual quality, and natural interactivity, striking a favorable balance between data costs and identity scalability.

cs.CV↗

arXiv2025

CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization

Multimodal large language models (MLLMs) rely heavily on instruction tuning to align vision and language capabilities, yet the computational cost of training on large-scale datasets remains a major bottleneck. Existing data selection methods aim to mitigate this by selecting important and diverse subsets, but they often suffer from two critical drawbacks: high computational overhead from processing the entire dataset and suboptimal data selection due to separate treatment of importance and diversity. We introduce CoIDO, a novel dual-objective framework that jointly optimizes data importance and diversity to overcome these challenges. Unlike existing approaches that require costly evaluations across the whole dataset, CoIDO employs a lightweight plug-in scorer. This scorer is trained on just a small random sample of data to learn the distribution of the candidate set, drastically reducing computational demands. By leveraging a homoscedastic uncertainty-based formulation, CoIDO effectively balances importance and diversity during training, enabling efficient and scalable data selection. In our experiments, we trained the CoIDO scorer using only 20 percent of randomly sampled data. Once trained, CoIDO was applied to the entire dataset to select a 20 percent subset for instruction tuning. On the widely used LLaVA-1.5-7B model across ten downstream tasks, this selected subset achieved an impressive 98.2 percent of the performance of full-data fine-tuning, on average.

cs.CV↗

arXiv2025

Conda: Column-Normalized Adam for Training Large Language Models Faster

Large language models (LLMs) have demonstrated impressive generalization and emergent capabilities, yet their pre-training remains computationally expensive and sensitive to optimization dynamics. While Adam-based optimizers offer fast convergence by adapting learning rates coordinate-wise, recent studies reveal that their updates often suffer from poor spectral conditioning and low-rank structures, hindering efficiency. Muon addresses this issue via global spectral normalization but lacks the per-coordinate adaptivity of Adam. In this work, we propose Column-Normalized Adam (Conda), a novel optimizer that bridges the strengths of both approaches. Conda projects updates into an orthogonal subspace and applies column-wise second moment normalization based on the projected gradients, thereby achieving both improved spectral conditioning and maintaining coordinate-wise adaptivity. This design alleviates the spectral pathologies of Adam while preserving its fast convergence behavior. Extensive experiments on the LLaMA and GPT-2 series show that Conda consistently outperforms AdamW, Muon, and other baselines in pre-training. Remarkably, on the LLaMA series, Conda achieves 2-2.5 the convergence speed of AdamW, measured in both training steps and training time. Further ablations demonstrate its robustness under diverse training setups. These results collectively highlight Conda as an effective and broadly applicable optimizer for large-scale LLM training. The code is released on https://github.com/jie040109/Conda

cs.LG↗

arXiv2025

A Chain-of-thought Reasoning Breast Ultrasound Dataset Covering All Histopathology Categories

Breast ultrasound (BUS) is an essential tool for diagnosing breast lesions, with millions of examinations per year. However, publicly available high-quality BUS benchmarks for AI development are limited in data scale and annotation richness. In this work, we present BUS-CoT, a BUS dataset for chain-of-thought (CoT) reasoning analysis, which contains 11,439 images of 10,019 lesions from 4,838 patients and covers all 99 histopathology types. To facilitate research on incentivizing CoT reasoning, we construct the reasoning processes based on observation, feature, diagnosis and pathology labels, annotated and verified by experienced experts. Moreover, by covering lesions of all histopathology types, we aim to facilitate robust AI systems in rare cases, which can be error-prone in clinical practice.

eess.IV↗

arXiv2025

SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning

Video large language models (Vid-LLMs) have shown strong capabilities in understanding video content. However, their reliance on dense video token representations introduces substantial memory and computational overhead in both prefilling and decoding. To mitigate the information loss of recent video token reduction methods and accelerate the decoding stage of Vid-LLMs losslessly, we introduce SpecVLM, a training-free speculative decoding (SD) framework tailored for Vid-LLMs that incorporates staged video token pruning. Building on our novel finding that the draft model's speculation exhibits low sensitivity to video token pruning, SpecVLM prunes up to 90% of video tokens to enable efficient speculation without sacrificing accuracy. To achieve this, we performs a two-stage pruning process: Stage I selects highly informative tokens guided by attention signals from the verifier (target model), while Stage II prunes remaining redundant ones in a spatially uniform manner. Extensive experiments on four video understanding benchmarks demonstrate the effectiveness and robustness of SpecVLM, which achieves up to 2.68$\times$ decoding speedup for LLaVA-OneVision-72B and 2.11$\times$ speedup for Qwen2.5-VL-32B. Code is available at https://github.com/zju-jiyicheng/SpecVLM.

cs.CV↗