Search arXiv⌕ Search

arXiv subjects

Le Zhou

Publications and source records attributed to Le Zhou.

At least 19 recordsLinked to original sources
arXiv2026

Breaking the Environment Wall: Evolving LLM Agent Environments for Recursive Self-Improvement

Many real-world tasks (e.g., office workflows, scientific experimentation) require LLM agents to interact repeatedly with their environments for context-dependent operations. However, such environments are often not agent-ready. First, information is often scattered and fragmented across the environment. Second, relevant evidence in the environment is often mixed with misleading information and conflicting versions. Third, environments evolve over time, introducing new noise and more challenging tasks. These challenges can substantially degrade performance for state-of-the-art AI agents (e.g., from 83.9% to 57.6%). To address these challenges, we propose Env-Rethink (a system with 27B post-trained model) that supports three main capabilities: (1) It adaptively builds Collection Maps (for organizing related files) and Event Logs (for contextualizing cross-data relationships) to supplement necessary context; (2) It further leverages the post-trained model (through offline trajectory learning) to identify underlying noise issues in the environment; (3) It ultimately evolves environments through virtual event histories that alter environmental states and evidence relationships, producing more tricky ones for further agent improvement. Experiments show that Env-Rethink can effectively improve downstream task performance (with over 15.1% rubric pass rate improvement across nine models on 30 tasks).

cs.AI↗
arXiv2026

The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement

Recursive self-improvement (RSI) enables AI systems to turn experience and feedback into persistent changes that improve both their capabilities and the process of future improvement. We first use the Headroom-Closed Index (HCI) to reveal the problems of existing LLMs, then introduce the RSI concept and its development roadmap: from improvement-execution autonomy, improvement-strategy autonomy, experience-acquisition autonomy, and environment-adaptation autonomy, to recursive meta-improvement. Next we examine RSI across scenarios (e.g., scientific discovery, embodied intelligence, software engineering), highlighting their distinct requirements and development speeds. Drawing on diverse industry practices and preliminary empirical evidence, we connect RSI research with practical systems and identify key challenges to achieving genuine RSI.

cs.LG↗
arXiv2026

Dial: A Knowledge-Grounded Dialect-Specific NL2SQL System

Enterprises commonly deploy heterogeneous database systems, each of which owns a distinct SQL dialect with different syntax rules, built-in functions, and execution constraints. However, most existing NL2SQL methods assume a single canonical dialect (e.g., SQLite) and struggle to produce queries that are both semantically correct and executable on target engines. Prompt-based approaches tightly couple intent reasoning with dialect syntax, rule-based translators often degrade native operators into generic constructs, and multi-dialect fine-tuning suffers from cross-dialect interference. In this paper, we present Dial, a knowledge-grounded framework for dialect-specific NL2SQL. Dial introduces: (1) a Dialect-Aware Logical Query Planning module that converts natural language into a dialect-aware logical query plan via operator-level intent decomposition and divergence-aware specification; (2) HINT-KB, a hierarchical intent-aware knowledge base that organizes dialect knowledge into i a Canonical Syntax Reference, ii a declarative function repository, and iii a procedural constraint repository; and (3) an execution-driven debugging and semantic verification loop that separates syntactic recovery from logic auditing to prevent semantic drift. We construct DS-NL2SQL, a benchmark covering six major database systems with 2,218 dialect-specific test cases. Experimental results show that Dial consistently improves translation accuracy by 10.25% and dialect feature coverage by 15.77% over state-of-the-art baselines.

cs.DB↗
arXiv2026

Spatial-sign-based multilinear principal component analysis for tensor data

Multilinear principal component analysis (MPCA) reduces the dimension of tensor-valued data while preserving their mode-specific structure, but its quadratic scatter criterion can be unstable under heavy-tailed distributions and contamination. We propose spatial-sign-based multilinear principal component analysis (SMPCA), a robust dimension-reduction method that centers the observations by their spatial median, removes radial magnitude through spatial-sign normalization, and estimates the mode-wise loading spaces by alternating eigendecompositions. Under a separable tensor elliptical model, we show that the target mode-wise loading spaces uniquely maximize the population criterion and that one complete sweep of exact population block updates recovers them from any initialization. We also characterize exactly when their tensor-product subspace coincides with a leading unrestricted subspace of vectorized spatial-sign PCA and, when finite second moments exist, ordinary vectorized PCA. At the sample level, we derive explicit statistical rates for the mode-wise subspaces and the joint multilinear projector, obtain corresponding reconstruction guarantees, establish consistency of the cumulative-contribution dimension selector, and prove that the objective values generated by exact cyclic updates are nondecreasing and convergent. Simulations and an empirical application show that SMPCA is more accurate and stable than competitors under heavy-tailed distributions and outlier contamination, while retaining competitive performance under light-tailed settings.

stat.ME↗
arXiv2026

High-Dimensional Change Point Analysis for Temporally Dependent Data

This paper develops adaptive procedures for detecting and locating mean changes in high-dimensional time series. Quadratic CUSUM statistics target dense changes, whereas coordinatewise maximum statistics target sparse changes. Two weighting schemes are considered to accommodate both interior and boundary changes. Under general non-Gaussian vector dependence, we establish the limiting distributions, validate the required centering and scaling estimators, and prove asymptotic independence between matched quadratic and maximum statistics. These results justify Cauchy combination tests. We further establish single-change localization and consistent multiple-change recovery using wild binary segmentation. Numerical results illustrate the effectiveness of the proposed methods.

stat.ME↗
arXiv2026

Bridging Modal Isolation in Interleaved Thinking: Supervising Modality Transitions via Stepwise Reinforcement

Interleaved thinking, where a unified multimodal model alternates between textual reasoning and visual generation, has shown promise on spatial and physical tasks. However, in complex long-chain scenarios, we identify a fundamental failure mode: generated images diverge from the textual context while subsequent text ignores the visual evidence, causing the two modalities to alternate without genuinely informing each other. We term this Modal Isolation and attribute it to compounding information loss at modality boundaries. We decompose each reasoning cycle into atomic operations and define modality transition loss, quantifying cross-modal hallucination (text-to-image) and visual utilization deficit (image-to-text) at each boundary. We propose MoTiF (Modality Tiransition Fidelity), a two-stage training framework that directly optimizes these transitions: Reflective SFT trains the model to detect and recover from erroneous visual outputs; Flow-GRPO improves image generation fidelity via reinforcement learning. All training signals in MoTiF derive from transition-level fidelity rather than end-task accuracy. Across four visual puzzle benchmarks, this transition-level supervision substantially improves both cross-modal coherence and final task accuracy. The results demonstrate that effective interleaved reasoning requires explicit structural supervision at modality boundaries, not merely scaling or end-task optimization.

cs.CV↗
arXiv2026

Elliptical Regularized Hotelling Testing for High Dimensional Data

We consider one-sample testing of a high-dimensional location parameter under elliptically symmetric distributions with heavy tails and pervasive cross-sectional dependence. We propose an elliptical regularized Hotelling test with Cauchy combination (ERHT--CC), based on the sample spatial median and the spatial-sign covariance matrix centered at that median. We derive its null asymptotic normality, consistent estimators of the centering and variance, and an explicit local power function. Since the power-optimal ridge parameter depends on the unknown alternative, we aggregate fixed-ridge $p$-values over a deterministic grid using the Cauchy rule. We establish a finite-grid joint Gaussian limit, justify the analytic combined $p$-value without estimating cross-ridge correlations, and characterize its local power. Simulation studies and an empirical analysis demonstrate the favorable finite-sample performance of ERHT--CC under heavy tails and pervasive dependence.

stat.ME↗
arXiv2026

Cauchy Aggregation of Ridge-Regularized Hotelling Tests for High-Dimensional Change-Point Detection

Ridge-regularized Hotelling-type (RHT) change-point tests depend on a ridge parameter $λ$, but the power-optimal value is determined by the unknown covariance structure and the unknown mean shift. We avoid selecting a single ridge value by computing fixed-ridge p-values on a finite deterministic grid and aggregating them with the Cauchy combination rule. Under the standard random-matrix conditions for fixed-ridge RHT statistics, we establish finite-grid joint weak convergence of the ridge processes. This leads to fixed-level validity under joint-limit calibration and small-tail validity for the analytic Cauchy p-value. Monte Carlo experiments show that deterministic-grid Cauchy aggregation has stable size behavior and achieves power close to the best stable fixed ridge choice across a range of covariance and signal configurations.

stat.ME↗
arXiv2026

Rose-SQL: Role-State Evolution Guided Structured Reasoning for Multi-Turn Text-to-SQL

Recent advances in Large Reasoning Models (LRMs) trained with Long Chain-of-Thought have demonstrated remarkable capabilities in code generation and mathematical reasoning. However, their potential in multi-turn Text-to-SQL tasks remains largely underexplored. Existing approaches typically rely on unstable API-based inference or require expensive fine-tuning on small-scale models. In this work, we present Rose-SQL, a training-free framework that leverages small-scale LRMs through in-context learning to enable accurate context-dependent parsing. We introduce the Role-State, a fine-grained representation that bridges the structural gap between schema linking and SQL generation by serving as a structural blueprint. To handle conversational dependencies, Rose-SQL traces the evolution of Role-State through historical context via structural isomorphism checks, guiding the model to infer the possible SQL composition for the current question through verified interaction trajectories. Experiments on the SParC and CoSQL benchmarks show that, within the Qwen3 series, Rose-SQL outperforms in-context learning baselines at the 4B scale and substantially surpasses state-of-the-art fine-tuned models at the 8B and 14B scales, while showing consistent gains on additional reasoning backbones.

cs.CL↗
arXiv2026

Rank-Based Sparse Regression in Principal Components Space under Measurement Error

We study high-dimensional regression in principal components space when the predictors are observed with additive measurement error and the response errors may be heavy-tailed. The starting point is the $\ell_1$-penalized principal-components estimator of Song and Zou (2026), which enjoys a blessing-of-dimensionality phenomenon under predictor contamination but senstive for heavy-tailed data or outliers. We replace the squared loss by a Wilcoxon-type rank loss and then apply a one-step adaptive reweighting scheme to reduce the shrinkage bias of the initial $\ell_1$ fit. The resulting procedure combines robustness to heavy-tailed response errors with the contamination geometry induced by the empirical principal-components basis. Our main theorem gives a prediction bound for the fixed-$λ$ second-stage fitted mean. Simulations show that the rank-based procedure is competitive under Gaussian noise and substantially more stable under heavy-tailed errors, especially when predictor contamination is present.

stat.ME↗
arXiv2026

A high-performance cobalt-free cathode for proton-conducting solid oxide fuel cells via multi-element doping in Sr2Fe2O6

The development of efficient and stable intermediate-temperature solid oxide fuel cells (SOFCs) necessitates high-performance cathode materials that are cobalt-free, cost-effective, and compatible with proton-conducting electrolytes. While Sr2Fe2O6 (SFO)-based ferrites offer a promising cobalt-free alternative, their electrochemical performance requires further enhancement to compete with state-of-the-art cathodes. This study proposes and validates a multi-element doping strategy as a superior approach to tailor the properties of SFO. The specific oxide Sr2Fe1.5Mo0.125Sn0.125Sc0.125Zr0.125O6 (SFO-ZSSM) is designed, synthesized via a solid-state reaction method, and systematically evaluated as a cathode for proton-conducting SOFCs (H-SOFCs). Its performance is benchmarked against a series of SFO cathodes modified with single dopants (Mo, Sn, Sc, Zr). Structural characterization confirms the successful formation of a phase-pure perovskite structure with homogeneous elemental distribution. Electrical conductivity relaxation (ECR) measurements reveal that SFO-ZSSM exhibits dramatically enhanced oxygen and proton transport kinetics compared to all singly-doped counterparts, demonstrating a significant synergistic effect. Consequently, fuel cells employing the SFO-ZSSM cathode deliver exceptional peak power densities of 1580, 1137, and 854 mW cm-2 at 700, 650, and 600 °C, respectively, significantly outperforming cells with single-doped cathodes. Electrochemical impedance spectroscopy further corroborates its superior catalytic activity, showing the lowest polarization resistance. Moreover, the SFO-ZSSM cell demonstrates excellent operational stability over 100 hours, attributed to its robust microstructure and Ba-free composition.

cond-mat.mtrl-sci↗
arXiv2025

Tensor Elliptical Graphic Model

We address the problem of robust estimation of sparse high dimensional tensor elliptical graphical model. Most of the research focus on tensor graphical model under normality. To extend the tensor graphical model to more heavy-tailed scenarios, motivated by the fact that up to a constant, the spatial-sign covariance matrix can approximate the true covariance matrix when the dimension turns to infinity under tensor elliptical distribution, we proposed a spatial-sign-based estimator to robustly estimate tensor elliptical graphical model, the rate of which matches the existing rate under normality for a wider family of distribution, i.e. elliptical distribution. We also conducted extensive simulations and real data applications to illustrate the practical utility of the proposed methods, especially under heavy-tailed distribution.

stat.ME↗
arXiv2025

ProDiff: Prototype-Guided Diffusion for Minimal Information Trajectory Imputation

Trajectory data is crucial for various applications but often suffers from incompleteness due to device limitations and diverse collection scenarios. Existing imputation methods rely on sparse trajectory or travel information, such as velocity, to infer missing points. However, these approaches assume that sparse trajectories retain essential behavioral patterns, which place significant demands on data acquisition and overlook the potential of large-scale human trajectory embeddings. To address this, we propose ProDiff, a trajectory imputation framework that uses only two endpoints as minimal information. It integrates prototype learning to embed human movement patterns and a denoising diffusion probabilistic model for robust spatiotemporal reconstruction. Joint training with a tailored loss function ensures effective imputation. ProDiff outperforms state-of-the-art methods, improving accuracy by 6.28\% on FourSquare and 2.52\% on WuXi. Further analysis shows a 0.927 correlation between generated and real trajectories, demonstrating the effectiveness of our approach.

cs.LG↗
arXiv2025

Why the hyperbolic polaritons are hyperbolic?

Polaritons travelling along a hyperbolic medium's surface have recently sparked significant interest in nanophotonics for the unprecedented manipulation ability on light at the nanoscale in a planar way, promising potential nano-optical applications, especially in two-dimensional circuitry. Despite of being named hyperbolic polaritons, the hyperbolic nature has not been thoroughly revealed since an analytical description of the Iso-frequency contour is still elusive. In this work, we proposed an analytical form for describing the iso-frequency contour of the hyperbolic polaritons, showcasing their strictly hyperbolic nature. Such an analytical form is obtained based on the focusing behavior of the hyperbolic polaritons and verified by both the published data from commonly used hyperbolic media systems of the hyperbolic polaritons and our own experimental characterizations on a hyperbolic metamaterial film. By presenting a concise and intuitive physical image, this work may provide a groundbreaking methodology in developing novel hyperbolic polaritons based optical devices.

physics.optics↗
arXiv2024

Machine Learning-Assisted 3D Printing of Thermoelectric Materials of Ultrahigh Performances at Room Temperature

Thermoelectric energy conversion is an attractive technology for generating electricity from waste heat and using electricity for solid-state cooling. However, conventional manufacturing processes for thermoelectric devices are costly and limited to simple device geometries. This work reports an extrusion printing method to fabricate high-performance thermoelectric materials with complex 3D architectures. By integrating high-throughput experimentation and Bayesian optimization (BO), our approach significantly accelerates the simultaneous search for the optimal ink formulation and printing parameters that deliver high thermoelectric performances while maintaining desired shape fidelity. A Gaussian process regression (GPR)-based machine learning model is employed to expeditiously predict thermoelectric power factor as a function of ink formulation and printing parameters. The printed bismuth antimony telluride (BiSbTe)-based thermoelectric materials under the optimized conditions exhibit an ultrahigh room temperature zT of 1.3, which is by far the highest in the printed thermoelectric materials. The machine learning-guided ink-based printing strategy can be highly generalizable to a wide range of functional materials and devices for broad technological applications.

physics.app-ph↗
arXiv2024

GeoGalactica: A Scientific Large Language Model in Geoscience

Large language models (LLMs) have achieved huge success for their general knowledge and ability to solve a wide spectrum of tasks in natural language processing (NLP). Due to their impressive abilities, LLMs have shed light on potential inter-discipline applications to foster scientific discoveries of a specific domain by using artificial intelligence (AI for science, AI4S). In the meantime, utilizing NLP techniques in geoscience research and practice is wide and convoluted, contributing from knowledge extraction and document classification to question answering and knowledge discovery. In this work, we take the initial step to leverage LLM for science, through a rather straightforward approach. We try to specialize an LLM into geoscience, by further pre-training the model with a vast amount of texts in geoscience, as well as supervised fine-tuning (SFT) the resulting model with our custom collected instruction tuning dataset. These efforts result in a model GeoGalactica consisting of 30 billion parameters. To our best knowledge, it is the largest language model for the geoscience domain. More specifically, GeoGalactica is from further pre-training of Galactica. We train GeoGalactica over a geoscience-related text corpus containing 65 billion tokens, preserving as the largest geoscience-specific text corpus. Then we fine-tune the model with 1 million pairs of instruction-tuning data consisting of questions that demand professional geoscience knowledge to answer. In this technical report, we will illustrate in detail all aspects of GeoGalactica, including data collection, data cleaning, base model selection, pre-training, SFT, and evaluation. We open-source our data curation tools and the checkpoints of GeoGalactica during the first 3/4 of pre-training.

cs.CL↗
arXiv2024

Fovea Transformer: Efficient Long-Context Modeling with Structured Fine-to-Coarse Attention

The quadratic complexity of self-attention in Transformers has hindered the processing of long text. To alleviate this problem, previous works have proposed to sparsify the attention matrix, taking advantage of the observation that crucial information about a token can be derived from its neighbors. These methods typically combine one or another form of local attention and global attention. Such combinations introduce abrupt changes in contextual granularity when going from local to global, which may be undesirable. We believe that a smoother transition could potentially enhance model's ability to capture long-context dependencies. In this study, we introduce Fovea Transformer, a long-context focused transformer that addresses the challenges of capturing global dependencies while maintaining computational efficiency. To achieve this, we construct a multi-scale tree from the input sequence, and use representations of context tokens with a progressively coarser granularity in the tree, as their distance to the query token increases. We evaluate our model on three long-context summarization tasks\footnote{Our code is publicly available at: \textit{https://github.com/ZiweiHe/Fovea-Transformer}}. It achieves state-of-the-art performance on two of them, and competitive results on the third with mixed improvement and setback of the evaluation metrics.

cs.CL↗
arXiv2023

Optical vortices enabled by structural vortices

The structural symmetry of solids plays an important role in defining their linear and nonlinear optical properties. The quest for versatile, cost-effective, large-scale, and defect-free approaches and materials platforms for tailoring structural and optical properties on demand has been underway for decades. We experimentally demonstrate a bottom-up self-assembly-based organic engineered material comprised of synthesized molecules with large dipole moments that are crystallized into a spherulite structure. The molecules align in an azimuthal direction, resulting in a vortex polarity with spontaneously broken symmetry leading to strong optical anisotropy and nonlinear optical responses. These unique polarization properties of the judiciously designed organic spherulite combined with the symmetry of structured optical beams enable a plethora of new linear and nonlinear light-matter interactions, including the generation of optical vortex beams with complex spin states and on-demand topological charges at the fundamental, doubled, and tripled frequencies. The results of this work are likely to enable numerous applications in areas such as high-dimensional quantum information processing, with large capacity and high security. The demonstrated spherulite crystals facilitate stand-alone micro-scale devices that rely on the unique micro-scale spontaneous vortex polarity that is likely to enable future applications for high-dimensional quantum information processing, spatiotemporal optical vortices, and a novel platform for optical manipulation and trapping.

physics.optics↗