arXiv subjects

Lei Cao

Publications and source records attributed to Lei Cao.

At least 37 records · Page 2Linked to original sources

arXiv2026

KramaBench: A Benchmark for AI Systems on Data-to-Insight Pipelines over Data Lakes

Discovering insights from a real-world data lake potentially containing unclean, semi-structured, and unstructured data requires a variety of data processing tasks, ranging from extraction and cleaning to integration, analysis, and modeling. This process often also demands domain knowledge and project-specific insight. While AI models have shown remarkable results in reasoning and code generation, their abilities to design and execute complex pipelines that solve these data-lake-to-insight challenges remain unclear. We introduce KramaBench which consists of 104 manually curated and solved challenges spanning 1700 files, 24 data sources, and 6 domains. KramaBench focuses on testing the end-to-end capabilities of AI systems to solve challenges which require automated orchestration of different data tasks. KramaBench also features a comprehensive evaluation framework assessing the pipeline design and individual data task implementation abilities of AI systems. We evaluate 8 LLMs using our single-agent reference framework DS-Guru, alongside both open- and closed-source single- and multi-agent systems, and find that while current agentic systems may handle isolated data-science tasks and generate plausible draft pipelines, they struggle with producing working end-to-end pipelines. On KramaBench, the best system reaches only 55% end-to-end accuracy in the full data-lake setting. Even with perfect retrieval, the accuracy tops out at 62%. Leading LLMs can identify up to 42% of important data tasks but can only fully implement 20% of individual data tasks. Our code, reference framework, and data are available at https://github.com/mitdbg/KramaBench.

cs.DB↗

arXiv2026

Enhanced Graphene-Water Thermal Transport via Edge Functionalization without Compromising In-Plane Thermal Conductivity

Interfacial thermal transport between graphene and water plays a critical role in a wide range of thermal and energy applications. Although chemical functionalization can significantly enhance graphene-water interfacial thermal conductance, it often degrades graphene's intrinsic in-plane phonon transport. In this work, we perform a systematic deep neural network molecular dynamics study comparing edge-functionalized graphene nanoribbons with surface-functionalized graphene in aqueous environments. We demonstrate that functionalizing only 10% of the ribbon edges with hydroxyl groups increases the graphene-water interfacial thermal conductance by more than eightfold, primarily due to strengthened interfacial interactions and improved wettability at the edges. In contrast to basal-plane oxidation, edge functionalization largely preserves in-plane thermal conductivity. Importantly, hydroxyl edge groups exert competing effects on phonon transport: they introduce additional boundary scattering that suppresses heat conduction, while simultaneously passivating dangling bonds at bare edges, thereby reducing phonon localization and edge-induced scattering. This competition leads to a non-monotonic dependence of in-plane thermal conductivity on edge functionalization ratio. These results establish edge functionalization as an effective strategy for enhancing graphene-water interfacial thermal transport without sacrificing intrinsic phonon transport properties.

cond-mat.mes-hall↗

arXiv2026

Active Label Cleaning for Reliable Detection of Electron Dense Deposits in Transmission Electron Microscopy Images

Automated detection of electron dense deposits (EDD) in glomerular disease is hindered by the scarcity of high-quality labeled data. While crowdsourcing reduces annotation cost, it introduces label noise. We propose an active label cleaning method to efficiently denoise crowdsourced datasets. Our approach uses active learning to select the most valuable noisy samples for expert re-annotation, building high-accuracy cleaning models. A Label Selection Module leverages discrepancies between crowdsourced labels and model predictions for both sample selection and instance-level noise grading. Experiments show our method achieves 67.18% AP\textsubscript{50} on a private dataset, an 18.83% improvement over training on noisy labels. This performance reaches 95.79% of that with full expert annotation while reducing annotation cost by 73.30%. The method provides a practical, cost-effective solution for developing reliable medical AI with limited expert resources.

cs.CV↗

arXiv2026

A Pathway to Sub-meV Detection of the Dark Universe: Robust Electron Avalanche in the PN junction at 10 mK

The search for light dark matter and cosmic primordial neutrinos necessitates detectors with sub-millielectronvolt (sub-meV) energy thresholds. While superconducting quantum sensors have approached this sensitivity, they often face significant challenges regarding readout complexity and scalability. To address these limitations, we propose a hybrid Superconductor-Insulator-P-N (S-I-P-N) detector architecture. This concept combines the high sensitivity of superconducting Cooper pair breaking with the massive intrinsic gain of semiconductor electron avalanches. A critical prerequisite for this scheme is operation at millikelvin (mK) temperatures, raising the critical fundamental question of whether silicon PN junctions can sustain avalanche multiplication in a regime where carrier freeze-out is severe. Here, we experimentally validate the critical semiconductor amplification stage of the proposed detector. We demonstrate that Silicon Photomultipliers (SiPMs) retain robust Geiger-mode avalanche capabilities at 10 mK. We report a single-photoelectron gain of order 10$^6$ and a dark count rate as low as 5~mHz/mm$^2$, 7 orders of magnitude lower than at room temperature. These results confirm the viability of high-gain semiconductor readout in the deep cryogenic regime, clearing the primary obstacle regarding the semiconductor component for the realization of scalable, sub-meV threshold S-I-P-N detectors.

physics.ins-det↗

arXiv2026

Fusion of Multiscale Features Via Centralized Sparse-attention Network for EEG Decoding

Electroencephalography (EEG) signal decoding is a key technology that translates brain activity into executable commands, laying the foundation for direct brain-machine interfacing and intelligent interaction. To address the inherent spatiotemporal heterogeneity of EEG signals, this paper proposes a multi-branch parallel architecture, where each temporal scale is equipped with an independent spatial feature extraction module. To further enhance multi-branch feature fusion, we propose a Fusion of Multiscale Features via Centralized Sparse-attention Network (EEG-CSANet), a centralized sparse-attention network. It employs a main-auxiliary branch architecture, where the main branch models core spatiotemporal patterns via multiscale self-attention, and the auxiliary branch facilitates efficient local interactions through sparse cross-attention. Experimental results show that EEG-CSANet achieves state-of-the-art (SOTA) performance across five public datasets (BCIC-IV-2A, BCIC-IV-2B, HGD, SEED, and SEED-VIG), with accuracies of 88.54%, 91.09%, 97.15%, 96.03%, and 90.56%, respectively. Such performance demonstrates its strong adaptability and robustness across various EEG decoding tasks. Moreover, extensive ablation studies are conducted to enhance the interpretability of EEG-CSANet. In the future, we hope that EEG-CSANet could serve as a promising baseline model in the field of EEG signal decoding. The source code is publicly available at: https://github.com/Xiangrui-Cai/EEG-CSANet

cs.LG↗

arXiv2025

Cross-modal ultra-scale learning with tri-modalities of renal biopsy images for glomerular multi-disease auxiliary diagnosis

Constructing a multi-modal automatic classification model based on three types of renal biopsy images can assist pathologists in glomerular multi-disease identification. However, the substantial scale difference between transmission electron microscopy (TEM) image features at the nanoscale and optical microscopy (OM) or immunofluorescence microscopy (IM) images at the microscale poses a challenge for existing multi-modal and multi-scale models in achieving effective feature fusion and improving classification accuracy. To address this issue, we propose a cross-modal ultra-scale learning network (CMUS-Net) for the auxiliary diagnosis of multiple glomerular diseases. CMUS-Net utilizes multiple ultrastructural information to bridge the scale difference between nanometer and micrometer images. Specifically, we introduce a sparse multi-instance learning module to aggregate features from TEM images. Furthermore, we design a cross-modal scale attention module to facilitate feature interaction, enhancing pathological semantic information. Finally, multiple loss functions are combined, allowing the model to weigh the importance among different modalities and achieve precise classification of glomerular diseases. Our method follows the conventional process of renal biopsy pathology diagnosis and, for the first time, performs automatic classification of multiple glomerular diseases including IgA nephropathy (IgAN), membranous nephropathy (MN), and lupus nephritis (LN) based on images from three modalities and two scales. On an in-house dataset, CMUS-Net achieves an ACC of 95.37+/-2.41%, an AUC of 99.05+/-0.53%, and an F1-score of 95.32+/-2.41%. Extensive experiments demonstrate that CMUS-Net outperforms other well-known multi-modal or multi-scale methods and show its generalization capability in staging MN. Code is available at https://github.com/SMU-GL-Group/MultiModal_lkx/tree/main.

cs.CV↗

arXiv2025

Pattern Forcing (0,1)-Matrices

We introduce two related notions of pattern enforcement in $(0,1)$-matrices: $Q$-forcing and strongly $Q$-forcing, which formalize distinct ways a fixed pattern $Q$ must appear within a larger matrix. A matrix is $Q$-forcing if every submatrix can realize $Q$ after turning any number of $1$-entries into $0$-entries, and strongly $Q$-forcing if every $1$-entry belongs to a copy of $Q$. For $Q$-forcing matrices, we establish the existence and uniqueness of extremal constructions minimizing the number of $1$-entries, characterize them using Young diagrams and corner functions, and derive explicit formulas and monotonicity results. For strongly $Q$-forcing matrices, we show that the minimum possible number of $0$-entries of an $m\times n$ strongly $Q$-forcing matrix is always $O(m+n)$, determine the maximum possible number of $1$-entries of an $n\times n$ strongly $P$-forcing matrix for every $2\times2$ and $3\times3$ permutation matrix, and identify symmetry classes with identical extremal behavior. We further propose a conjectural formula for the maximum possible number of $1$-entries of an $n\times n$ strongly $I_k$-forcing matrix, supported by results for $k=2,3$. These findings reveal contrasting extremal structures between forcing and strongly forcing, extending the combinatorial understanding of pattern embedding in $(0,1)$-matrices.

math.CO↗

arXiv2025

Unstructured Data Analysis using LLMs: A Comprehensive Benchmark

Nowadays, the explosion of unstructured data presents immense analytical value. Leveraging the remarkable capability of large language models (LLMs) in extracting attributes of structured tables from unstructured data, researchers are developing LLM-powered data systems for users to analyze unstructured documents as working with a database. These unstructured data analysis (UDA) systems differ significantly in all aspects, including query interfaces, query optimization strategies, and operator implementations, making it unclear which performs best in which scenario. Unfortunately, there does not exist a comprehensive benchmark that offers high-quality, large-volume, and diverse datasets as well as rich query workload to thoroughly evaluate such systems. To fill this gap, we present UDA-Bench, the first benchmark for unstructured data analysis that meets all the above requirements. Specifically, we organize a team with 30 graduate students that spends over in total 10,000 hours on curating 5 datasets from various domains and constructing a relational database view from these datasets by manual annotation. These relational databases can be used as ground truth to evaluate any of these UDA systems despite their differences in programming interfaces. Moreover, we design diverse queries to analyze the attributes defined in the database schema, covering different types of analytical operators with varying selectivities and complexities. We conduct in-depth analysis of the key building blocks of existing UDA systems: query interface, query optimization, operator design, and data processing. We run exhaustive experiments over the benchmark to fully evaluate these systems and different techniques w.r.t. the above building blocks.

cs.DB↗

arXiv2025

Not All Instances Are Equally Valuable: Towards Influence-Weighted Dataset Distillation

Dataset distillation condenses large datasets into synthetic subsets, achieving performance comparable to training on the full dataset while substantially reducing storage and computation costs. Most existing dataset distillation methods assume that all real instances contribute equally to the process. In practice, real-world datasets contain both informative and redundant or even harmful instances, and directly distilling the full dataset without considering data quality can degrade model performance. In this work, we present Influence-Weighted Distillation IWD, a principled framework that leverages influence functions to explicitly account for data quality in the distillation process. IWD assigns adaptive weights to each instance based on its estimated impact on the distillation objective, prioritizing beneficial data while downweighting less useful or harmful ones. Owing to its modular design, IWD can be seamlessly integrated into diverse dataset distillation frameworks. Our empirical results suggest that integrating IWD tends to improve the quality of distilled datasets and enhance model performance, with accuracy gains of up to 7.8%.

cs.LG↗

arXiv2025

Handling Label Noise via Instance-Level Difficulty Modeling and Dynamic Optimization

Recent studies indicate that deep neural networks degrade in generalization performance under noisy supervision. Existing methods focus on isolating clean subsets or correcting noisy labels, facing limitations such as high computational costs, heavy hyperparameter tuning process, and coarse-grained optimization. To address these challenges, we propose a novel two-stage noisy learning framework that enables instance-level optimization through a dynamically weighted loss function, avoiding hyperparameter tuning. To obtain stable and accurate information about noise modeling, we introduce a simple yet effective metric, termed wrong event, which dynamically models the cleanliness and difficulty of individual samples while maintaining computational costs. Our framework first collects wrong event information and builds a strong base model. Then we perform noise-robust training on the base model, using a probabilistic model to handle the wrong event information of samples. Experiments on five synthetic and real-world LNL benchmarks demonstrate our method surpasses state-of-the-art methods in performance, achieves a nearly 75% reduction in computational time and improves model scalability.

cs.LG↗

arXiv2025

Characterization of SiPMs at 40 K for neutrino coherent detection based on pure CsI

Silicon photomultiplier (SiPM), as the core photoelectric sensor for coherent neutrino detection in low-temperature pure CsI, its working performance directly determines the measurement accuracy of the scintillator light yield. Our previous research has fully demonstrated the performance of pure CsI at liquid nitrogen temperature. More intriguingly, its performance is expected to be even better at 40 K. However, the performance characteristics of SiPM in the 40 K temperature range still remain to be explored. In this study, a self-developed adjustable temperature control system ranging from 30 K to 293 K was built to investigate the key performance parameters of SiPM at different temperatures, such as single photoelectron spectrum, gain, breakdown voltage, dark count rate, after-pulse, internal crosstalk, and single photoelectron resolution. Special emphasis was placed on examining the key performance parameters of SiPM in the 40 K temperature range to evaluate its feasibility for light yield measurement in this temperature range. The results show that this study obtained the parameter variation trends and optimal working conditions of 3 types of SiPM at different temperatures, thereby improving the sensitivity of the detector. This research provides important technical support for low-temperature detection in neutrino physics experiments.

physics.ins-det↗

arXiv2025

Glo-UMF: A Unified Multi-model Framework for Automated Morphometry of Glomerular Ultrastructural Characterization

Background and Objective: To address the inability of single-model architectures to perform simultaneous analysis of complex glomerular ultrastructures, we developed Glo-UMF, a unified multi-model framework integrating segmentation, classification, and detection to systematically quantify key ultrastructural features. Methods: Glo-UMF decouples quantification tasks by constructing three dedicated deep models: an ultrastructure segmentation model, a glomerular filtration barrier (GFB) region classification model, and an electron-dense deposits (EDD) detection model. Their outputs are integrated through a post-processing workflow with adaptive GFB cropping and measurement location screening, enhancing measurement reliability and providing comprehensive quantitative results that overcome the limitations of traditional grading. Results: Trained on 372 electron microscopy images, Glo-UMF enables simultaneous quantification of glomerular basement membrane (GBM) thickness, the degree of foot process effacement (FPE), and EDD location. In 115 test cases spanning 9 renal pathological types, the automated quantification results showed strong agreement with pathological reports, with an average processing time of 4.23$\pm$0.48 seconds per case on a CPU environment. Conclusions: The modular design of Glo-UMF allows for flexible extensibility, supporting the joint quantification of multiple features. This framework ensures robust generalization and clinical applicability, demonstrating significant potential as an efficient auxiliary tool in glomerular pathological analysis.

cs.CV↗

arXiv2025

Topotactic phase transition in epitaxial La0.7Sr0.3MnO3-δ films induced by oxygen getter assisted thermal annealing

Oxygen vacancies play a crucial role in controlling the physical properties of complex oxides. In La0.7Sr0.3MnO3-δ, the topotactic phase transition from Perovskite (PV) to Brownmillerite (BM) can be triggered e.g. via oxygen removal during thermal annealing. Here we report on a very efficient thermal vacuum annealing method using aluminum as an oxygen getter material. The topotactic phase transition is characterized by X-ray Diffraction which confirms a successful transition from PV to BM in La0.7Sr0.3MnO3-δ thin films grown via physical vapor deposition. The efficiency of this method is confirmed using La0.7Sr0.3MnO3-δ micron-sized bulk powder. The accompanying transition from the original Ferromagnetic (FM) to an Antiferromagnetic (AF) state and the simultaneous transition from a metallic to an insulating state is characterized using Superconducting Quantum Interference Device (SQUID)-magnetometry and Alternating Current (AC) resistivity measurements, respectively. The near surface manganese oxidation states are probed by synchrotron X-ray Absorption Spectroscopy. Moreover, X-ray Reflectivity, Atomic Force Microscopy and Scanning Transmission Electron Microscopy reveal surface segregation and cation redistribution during the oxygen getter assisted annealing process.

cond-mat.mtrl-sci↗

arXiv2025

Leaky Forcing: Extending Zero Forcing Results to a Fault-Tolerant Setting

We study a recent variation of zero forcing called leaky forcing. Zero forcing is a propagation process on a network whereby some nodes are initially blue with all others white. Blue vertices can "force" a white neighbor to become blue if all other neighbors are blue. The goal is to find the minimum number of initially blue vertices to eventually force all vertices blue after exhaustively applying the forcing rule above. Leaky forcing is a fault-tolerant variation of zero forcing where certain vertices (not necessarily initially blue) cannot force. The goal in this context is to find the minimum number of initially blue vertices needed that can eventually force all vertices to be blue, regardless of which small number of vertices can't force. This work extends results from zero forcing in terms of leaky forcing. In particular, we provide a complete determination of leaky forcing numbers for all unicyclic graphs and upper bounds for generalized Petersen graphs. We also provide bounds for the effect of both edge removal and vertex removal on the $\ell$-leaky forcing number. Finally, we completely characterize connected graphs that have the minimum and maximum possible $1$-leaky forcing number (i.e., when $Z_{1}(G) = 2$ and when $Z_{1}(G) = |V(G)|-1$).

math.CO↗

arXiv2025

QUEST: Query Optimization in Unstructured Document Analysis

Most recently, researchers have started building large language models (LLMs) powered data systems that allow users to analyze unstructured text documents like working with a database because LLMs are very effective in extracting attributes from documents. In such systems, LLM-based extraction operations constitute the performance bottleneck of query execution due to the high monetary cost and slow LLM inference. Existing systems typically borrow the query optimization principles popular in relational databases to produce query execution plans, which unfortunately are ineffective in minimizing LLM cost. To fill this gap, we propose QUEST, which features a bunch of novel optimization strategies for unstructured document analysis. First, we introduce an index-based strategy to minimize the cost of each extraction operation. With this index, QUEST quickly retrieves the text segments relevant to the target attributes and only feeds them to LLMs. Furthermore, we design an evidence-augmented retrieval strategy to reduce the possibility of missing relevant segments. Moreover, we develop an instance-optimized query execution strategy: because the attribute extraction cost could vary significantly document by document, QUEST produces different plans for different documents. For each document, QUEST produces a plan to minimize the frequency of attribute extraction. The innovations include LLM cost-aware operator ordering strategies and an optimized join execution approach that transforms joins into filters. Extensive experiments on 3 real-world datasets demonstrate the superiority of QUEST, achieving 30%-6x cost savings while improving the F1 score by 10% -27% compared with state-of-the-art baselines.

cs.DB↗

arXiv2025

AutoPrep: Natural Language Question-Aware Data Preparation with a Multi-Agent Framework

Answering natural language (NL) questions about tables, known as Tabular Question Answering (TQA), is crucial because it allows users to quickly and efficiently extract meaningful insights from structured data, effectively bridging the gap between human language and machine-readable formats. Many of these tables are derived from web sources or real-world scenarios, which require meticulous data preparation (or data prep) to ensure accurate responses. However, preparing such tables for NL questions introduces new requirements that extend beyond traditional data preparation. This question-ware data preparation involves specific tasks such as column derivation and filtering tailored to particular questions, as well as question-aware value normalization or conversion, highlighting the need for a more nuanced approach in this context. Because each of the above tasks is unique, a single model (or agent) may not perform effectively across all scenarios. In this paper, we propose AutoPrep, a large language model (LLM)-based multiagent framework that leverages the strengths of multiple agents, each specialized in a certain type of data prep, ensuring more accurate and contextually relevant responses. Given an NL question over a table, AutoPrep performs data prep through three key components. Planner: Determines a logical plan, outlining a sequence of high-level operations. Programmer: Translates this logical plan into a physical plan by generating the corresponding low-level code. Executor: Executes the generated code to process the table. To support this multi-agent framework, we design a novel Chain-ofClauses reasoning mechanism for high-level operation suggestion, and a tool-augmented method for low-level code generation.

cs.CL↗

arXiv2025

Tensor-based Graph Learning with Consistency and Specificity for Multi-view Clustering

In the context of multi-view clustering, graph learning is recognized as a crucial technique, which generally involves constructing an adaptive neighbor graph based on probabilistic neighbors, and then learning a consensus graph for clustering. However, it is worth noting that these graph learning methods encounter two significant limitations. Firstly, they often rely on Euclidean distance to measure similarity when constructing the adaptive neighbor graph, which proves inadequate in capturing the intrinsic structure among data points in practice, particularly for high-dimensional data. Secondly, most of these methods focus solely on consensus graph, ignoring unique information from each view. Although a few graph-based studies have considered using specific information as well, the modelling approach employed does not exclude the noise impact from the common or specific components. To this end, we propose a novel tensor-based multi-view graph learning framework that simultaneously considers consistency and specificity, while effectively eliminating the influence of noise. Specifically, we calculate similarity using pseudo-Stiefel manifold distance to preserve the intrinsic properties of data. By making an assumption that the learned neighbor graph of each view comprises a consistent part, a specific part, and a noise part, we formulate a new tensor-based target graph learning paradigm for noise-free graph fusion. Owing to the benefits of tensor singular value decomposition (t-SVD) in uncovering high-order correlations, this model is capable of achieving a comprehensive understanding of the target graph. Furthermore, we derive an algorithm to address the optimization problem. Experiments on six datasets have demonstrated the superiority of our method. We have released the source code on https://github.com/lshi91/CSTGL-Code.

cs.LG↗

arXiv2025

HARMONY: A Scalable Distributed Vector Database for High-Throughput Approximate Nearest Neighbor Search

Approximate Nearest Neighbor Search (ANNS) is essential for various data-intensive applications, including recommendation systems, image retrieval, and machine learning. Scaling ANNS to handle billions of high-dimensional vectors on a single machine presents significant challenges in memory capacity and processing efficiency. To address these challenges, distributed vector databases leverage multiple nodes for the parallel storage and processing of vectors. However, existing solutions often suffer from load imbalance and high communication overhead, primarily due to traditional partition strategies that fail to effectively distribute the workload. In this paper, we introduce Harmony, a distributed ANNS system that employs a novel multi-granularity partition strategy, combining dimension-based and vector-based partition. This strategy ensures a balanced distribution of computational load across all nodes while effectively minimizing communication costs. Furthermore, Harmony incorporates an early-stop pruning mechanism that leverages the monotonicity of distance computations in dimension-based partition, resulting in significant reductions in both computational and communication overhead. We conducted extensive experiments on diverse real-world datasets, demonstrating that Harmony outperforms leading distributed vector databases, achieving 4.63 times throughput on average in four nodes and 58% performance improvement over traditional distribution for skewed workloads.

cs.DB↗