arXiv subjects

Jianfeng Zhang

Publications and source records attributed to Jianfeng Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

CauKer: Classification Time Series Foundation Models Can Be Pretrained on Synthetic Data

Time series foundation models (TSFMs) have recently gained significant attention due to their strong zero-shot capabilities and widespread real-world applications. Such models typically require a computationally costly pre-training on large-scale, carefully curated collections of real-world sequences. To allow for a sample-efficient pre-training of TSFMs, we propose \textsc{CauKer}, a novel algorithm designed to generate diverse, causally coherent synthetic time series with realistic trends, seasonality, and nonlinear interactions. \textsc{CauKer} combines Gaussian Process (GP) kernel composition with Structural Causal Models (SCM) to produce data for sample-efficient pre-training of state-of-the-art classification TSFMs having different architectures and following different pre-training approaches. Additionally, our experiments reveal that \textsc{CauKer}-generated datasets exhibit clear scaling laws for both dataset size (10K to 10M samples) and model capacity (1M to 783M parameters), unlike real-world datasets, which display irregular scaling behavior. The source code is publicly available at https://github.com/ShifengXIE/CauKer.

cs.LG↗

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

MantisV2: Closing the Zero-Shot Gap in Time Series Classification with Synthetic Data and Test-Time Strategies

Developing foundation models for time series classification is of high practical relevance, as such models can serve as universal feature extractors for diverse downstream tasks. Although early models such as Mantis have shown the promise of this approach, a substantial performance gap remained between frozen and fine-tuned encoders. In this work, we introduce methods that significantly strengthen zero-shot feature extraction for time series. First, we introduce Mantis+, a variant of Mantis pre-trained entirely on synthetic time series. Second, through controlled ablation studies, we refine the architecture and obtain MantisV2, an improved and more lightweight encoder. Third, we propose an enhanced test-time methodology that leverages intermediate-layer representations and refines output-token aggregation. In addition, we show that performance can be further improved via self-ensembling and cross-model embedding fusion. Extensive experiments on UCR, UEA, Human Activity Recognition (HAR) benchmarks, and EEG datasets show that MantisV2 and Mantis+ consistently outperform prior time series foundation models, achieving state-of-the-art zero-shot performance.

cs.LG↗

arXiv2026

Cognitive Chunking for Soft Prompts: Accelerating Compressor Learning via Block-wise Causal Masking

Providing extensive context via prompting is vital for leveraging the capabilities of Large Language Models (LLMs). However, lengthy contexts significantly increase inference latency, as the computational cost of self-attention grows quadratically with sequence length. To mitigate this issue, context compression-particularly soft prompt compressio-has emerged as a widely studied solution, which converts long contexts into shorter memory embeddings via a trained compressor. Existing methods typically compress the entire context indiscriminately into a set of memory tokens, requiring the compressor to capture global dependencies and necessitating extensive pre-training data to learn effective patterns. Inspired by the chunking mechanism in human working memory and empirical observations of the spatial specialization of memory embeddings relative to original tokens, we propose Parallelized Iterative Compression (PIC). By simply modifying the Transformer's attention mask, PIC explicitly restricts the receptive field of memory tokens to sequential local chunks, thereby lowering the difficulty of compressor training. Experiments across multiple downstream tasks demonstrate that PIC consistently outperforms competitive baselines, with superiority being particularly pronounced in high compression scenarios (e.g., achieving relative improvements of 29.8\% in F1 score and 40.7\% in EM score on QA tasks at the $64\times$ compression ratio). Furthermore, PIC significantly expedites the training process. Specifically, when training the 16$\times$ compressor, it surpasses the peak performance of the competitive baseline while effectively reducing the training time by approximately 40\%.

cs.AI↗

arXiv2025

Extreme Strain Controlled Correlated Metal-Insulator Transition in the Altermagnet CrSb

Correlated flat bands and altermagnetism are two important directions in quantum materials, centred respectively on interaction-dominated phases and symmetry-enforced spin-textured states, yet both derive from lattice symmetry and orbital hybridization. This common origin implies that extreme crystal distortion, by narrowing bandwidths, enhancing correlations and reshaping the symmetries of altermagnetic spin splittings, could unify flat-band and altermagnetic physics in a single material; in practice, however, achieving such large distortions in a crystalline altermagnet is a formidable challenge. Here we combine a dedicated strain device with a tailored single-crystal mounting scheme to impose a highly tensile strain gradient in bulk CrSb, a prototypical altermagnet, creating a near-surface layer in which the in-plane lattice is strongly distorted relative to the weakly strained bulk, while the average bulk distortion remains small. Angle-resolved photoemission reveals a reversible regime at moderate strain, where a deeper flat-band feature, attributed to a strain-gradient-driven suppression of Cr-Sb hybridization, coexists with a correlation-enhanced Cr 3d flat band, and an irreversible regime at larger strain where partial bond decoupling drives a predominantly insulating spectral response. Density-functional calculations show that an orbital-selective altermagnetic spin texture persists across this correlated regime despite strong bandwidth renormalisation. These results define a strain-symmetry-correlation map for CrSb and establish extreme tensile strain as a route to co-engineer flat-band tendencies and spin-textured, zero-net-moment correlated states in altermagnets, pointing toward strain-adaptive, spin-selective Mott filtering and related device concepts.

cond-mat.str-el↗

arXiv2025

Hidden Moiré Topology of Low-Symmetry Weyl Surfaces

Topological materials are defined by the correspondence between bulk topology and boundary states, yet this correspondence becomes enigmatic on low-symmetry surfaces where bulk and surface periodicities are inherently mismatched. Here we reveal a hidden moiré topology emerging on the (103) surface of the Weyl semimetal NdAlSi. Angle-resolved photoemission spectroscopy uncovers closed Fermi-arc loops and momentum-space moiré modulations, phenomena unanticipated in conventional topological theory. We show that these emerge from incomplete bulk projection and multi-cell interference governed by a least-common-multiple framework. Least-common-multiple guided DFT and Green's-function calculations quantitatively reproduce the observed spectra, establishing the universality of this commensuration rule. These findings transform a long-standing paradox of bulk-boundary correspondence into a new paradigm of momentum-space moiré reconstruction, bridging crystalline and quasicrystalline topologies and opening routes to flat-band engineering on complex surfaces.

cond-mat.mtrl-sci↗

arXiv2025

Wealth or Stealth? The Camouflage Effect in Insider Trading

We consider a Kyle-type model where insider trading takes place among a potentially large population of liquidity traders and is subject to legal penalties. Insiders exploit the liquidity provided by the trading masses to "camouflage" their actions and balance expected wealth with the necessary stealth to avoid detection. Under a diverse spectrum of prosecution schemes, we establish the existence of equilibria for arbitrary population sizes and a unique limiting equilibrium. A convergence analysis determines the scale of insider trading by a stealth index $γ$, revealing that the equilibrium can be closely approximated by a simple limit due to diminished price informativeness. Empirical aspects are derived from two calibration experiments using non-overlapping data sets spanning from 1980 to 2018, which underline the indispensable role of a large population in insider trading models with legal risk, along with important implications for the incidence of stealth trading and the deterrent effect of legal enforcement.

econ.GN↗

arXiv2025

Automatic nodule identification and differentiation in ultrasound videos to facilitate per-nodule examination

Ultrasound is a vital diagnostic technique in health screening, with the advantages of non-invasive, cost-effective, and radiation free, and therefore is widely applied in the diagnosis of nodules. However, it relies heavily on the expertise and clinical experience of the sonographer. In ultrasound images, a single nodule might present heterogeneous appearances in different cross-sectional views which makes it hard to perform per-nodule examination. Sonographers usually discriminate different nodules by examining the nodule features and the surrounding structures like gland and duct, which is cumbersome and time-consuming. To address this problem, we collected hundreds of breast ultrasound videos and built a nodule reidentification system that consists of two parts: an extractor based on the deep learning model that can extract feature vectors from the input video clips and a real-time clustering algorithm that automatically groups feature vectors by nodules. The system obtains satisfactory results and exhibits the capability to differentiate ultrasound videos. As far as we know, it's the first attempt to apply re-identification technique in the ultrasonic field.

eess.IV↗

arXiv2025

Active Contour Models Driven by Hyperbolic Mean Curvature Flow for Image Segmentation

Parabolic mean curvature flow-driven active contour models (PMCF-ACMs) are widely used for image segmentation, yet they suffer severe degradation under high-intensity noise because gradient-descent evolutions exhibit the well-known zig-zag phenomenon. To overcome this drawback, we propose hyperbolic mean curvature flow-driven ACMs (HMCF-ACMs). This novel framework incorporates an adjustable acceleration field to autonomously regulate curve evolution smoothness, providing dual degrees of freedom for adaptive selection of both initial contours and velocity fields. We rigorously prove that HMCF-ACMs are normal flows and establish their numerical equivalence to wave equations through a level set formulation with signed distance functions. An efficient numerical scheme combining spectral discretization and optimized temporal integration is developed to solve the governing equations, and its stability condition is derived through Fourier analysis. Extensive experiments on natural and medical images validate that HMCF-ACMs achieve superior performance under high-noise conditions, demonstrating reduced parameter sensitivity, enhanced noise robustness, and improved segmentation accuracy compared to PMCF-ACMs.

cs.CV↗

arXiv2025

Seed3D 1.0: From Images to High-Fidelity Simulation-Ready 3D Assets

Developing embodied AI agents requires scalable training environments that balance content diversity with physics accuracy. World simulators provide such environments but face distinct limitations: video-based methods generate diverse content but lack real-time physics feedback for interactive learning, while physics-based engines provide accurate dynamics but face scalability limitations from costly manual asset creation. We present Seed3D 1.0, a foundation model that generates simulation-ready 3D assets from single images, addressing the scalability challenge while maintaining physics rigor. Unlike existing 3D generation models, our system produces assets with accurate geometry, well-aligned textures, and realistic physically-based materials. These assets can be directly integrated into physics engines with minimal configuration, enabling deployment in robotic manipulation and simulation training. Beyond individual objects, the system scales to complete scene generation through assembling objects into coherent environments. By enabling scalable simulation-ready content creation, Seed3D 1.0 provides a foundation for advancing physics-based world simulators. Seed3D 1.0 is now available on https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?modelId=doubao-seed3d-1-0-250928&tab=Gen3D

eess.IV↗

arXiv2025

Characterizing nonconvex boundaries via scalarization

We present a unified approach for characterizing the boundary of a possibly nonconvex domain. Motivated by the well-known Pascoletti--Serafini method of scalarization, we recast the boundary characterization as a multi-criteria optimization problem with respect to a local partial order induced by a spherical cone with varying orient. Such an approach enables us to trace the whole boundary and can be considered a general dual representation for arbitrary (nonconvex) sets satisfying an exterior cone condition. We prove the equivalence between the geometrical boundary and the scalarization-implied boundary, particularly in the case of Euclidean spaces and two infinite-dimensional spaces for practical interest. By reformulating each scalarized problem as a parameterized constrained optimization problem, we shall develop a corresponding numerical scheme for the proposed approach. Some related applications are also discussed.

math.OC↗

arXiv2025

Stand on The Shoulders of Giants: Building JailExpert from Previous Attack Experience

Large language models (LLMs) generate human-aligned content under certain safety constraints. However, the current known technique ``jailbreak prompt'' can circumvent safety-aligned measures and induce LLMs to output malicious content. Research on Jailbreaking can help identify vulnerabilities in LLMs and guide the development of robust security frameworks. To circumvent the issue of attack templates becoming obsolete as models evolve, existing methods adopt iterative mutation and dynamic optimization to facilitate more automated jailbreak attacks. However, these methods face two challenges: inefficiency and repetitive optimization, as they overlook the value of past attack experiences. To better integrate past attack experiences to assist current jailbreak attempts, we propose the \textbf{JailExpert}, an automated jailbreak framework, which is the first to achieve a formal representation of experience structure, group experiences based on semantic drift, and support the dynamic updating of the experience pool. Extensive experiments demonstrate that JailExpert significantly improves both attack effectiveness and efficiency. Compared to the current state-of-the-art black-box jailbreak methods, JailExpert achieves an average increase of 17\% in attack success rate and 2.7 times improvement in attack efficiency. Our implementation is available at \href{https://github.com/xiZAIzai/JailExpert}{XiZaiZai/JailExpert}

cs.CR↗

arXiv2025

Instability and Efficiency of Non-cooperative Games

It is well known that a non-cooperative game may have multiple equilibria. In this paper we consider the efficiency of games, measured by the ratio between the aggregate payoff over all Nash equilibria and that over all admissible controls. Such efficiency operator is typically unstable with respect to small perturbation of the game. This seemingly bad property can actually be a good news in practice: it is possible that a small change of the game mechanism may improve the efficiency of the game dramatically. We shall introduce a game mediator with limited resources and investigate two mechanism designs aiming to improve the efficiency. Moreover, we compare the mediator's capability of efficiency improvement when she has access to full information or only partial information. When the mechanisms contain only rewards, the mediator has the same power in the two cases. However, when the mediator can use punishments as well, in general she may have a larger power to improve the efficiency in the full information case than in the partial information case.

cs.GT↗

arXiv2025

Set-valued Hamilton-Jacobi-Bellman Equations

Building upon the dynamic programming principle for set-valued functions arising from many applications, in this paper we propose a new notion of set-valued PDEs. The key component in the theory is a set-valued Itô formula, characterizing the flows on the surface of the dynamic sets. In the contexts of multivariate control problems, we establish the wellposedness of the set-valued HJB equations, which extends the standard HJB equations in the scalar case to the multivariate case. As an application, a moving scalarization for certain time inconsistent problems is constructed by using the classical solution of the set-valued HJB equation.

math.OC↗

arXiv2025

Puppeteer: Rig and Animate Your 3D Models

Modern interactive applications increasingly demand dynamic 3D content, yet the transformation of static 3D models into animated assets constitutes a significant bottleneck in content creation pipelines. While recent advances in generative AI have revolutionized static 3D model creation, rigging and animation continue to depend heavily on expert intervention. We present Puppeteer, a comprehensive framework that addresses both automatic rigging and animation for diverse 3D objects. Our system first predicts plausible skeletal structures via an auto-regressive transformer that introduces a joint-based tokenization strategy for compact representation and a hierarchical ordering methodology with stochastic perturbation that enhances bidirectional learning capabilities. It then infers skinning weights via an attention-based architecture incorporating topology-aware joint attention that explicitly encodes inter-joint relationships based on skeletal graph distances. Finally, we complement these rigging advances with a differentiable optimization-based animation pipeline that generates stable, high-fidelity animations while being computationally more efficient than existing approaches. Extensive evaluations across multiple benchmarks demonstrate that our method significantly outperforms state-of-the-art techniques in both skeletal prediction accuracy and skinning quality. The system robustly processes diverse 3D content, ranging from professionally designed game assets to AI-generated shapes, producing temporally coherent animations that eliminate the jittering issues common in existing methods.

cs.CV↗

arXiv2025

Viscosity Solutions for HJB Equations on the Process Space

In this paper we investigate a path dependent optimal control problem on the process space with both drift and volatility controls, with possibly degenerate volatility. The dynamic value function is characterized by a fully nonlinear second order path dependent HJB equation on the process space, which is by nature infinite dimensional. In particular, our model covers mean field control problems with common noise as a special case. We shall introduce a new notion of viscosity solutions and establish both the existence and the comparison principle, under merely Lipschitz/Holder continuity assumptions. The main feature of our notion is that, besides the standard smooth part, the test function consists of an extra singular component which allows us to handle the second order derivatives of the smooth test functions without invoking the Crandall-Ishii lemma. We shall use the doubling variable arguments, combined with the Ekeland-Borwein-Preiss variational principle in order to overcome the noncompactness of the state space. A smooth gauge-type function on the path space is crucial for our estimates.

math.OC↗

arXiv2025

SMPL Normal Map Is All You Need for Single-view Textured Human Reconstruction

Single-view textured human reconstruction aims to reconstruct a clothed 3D digital human by inputting a monocular 2D image. Existing approaches include feed-forward methods, limited by scarce 3D human data, and diffusion-based methods, prone to erroneous 2D hallucinations. To address these issues, we propose a novel SMPL normal map Equipped 3D Human Reconstruction (SEHR) framework, integrating a pretrained large 3D reconstruction model with human geometry prior. SEHR performs single-view human reconstruction without using a preset diffusion model in one forward propagation. Concretely, SEHR consists of two key components: SMPL Normal Map Guidance (SNMG) and SMPL Normal Map Constraint (SNMC). SNMG incorporates SMPL normal maps into an auxiliary network to provide improved body shape guidance. SNMC enhances invisible body parts by constraining the model to predict an extra SMPL normal Gaussians. Extensive experiments on two benchmark datasets demonstrate that SEHR outperforms existing state-of-the-art methods.

cs.CV↗

arXiv2025

A New Approach for the Continuous Time Kyle-Back Strategic Insider Equilibrium Problem

This paper considers a continuous time Kyle-Back model which is a game problem between an insider and a market marker. The existing literature typically focuses on the existence of equilibrium by using the PDE approach, which requires certain Markovian structure and the equilibrium is in the bridge form. We shall provide a new approach which is used widely for stochastic controls and stochastic differential games. We characterize all equilibria through a coupled system of forward backward SDEs, where the forward one is the conditional law of the inside information and the backward one is the insider's optimal value. In particular, when the time duration is small, we show that the FBSDE is wellposed and thus the game has a unique equilibrium. This is the first uniqueness result in the literature, without restricting the equilibria to certain special structure. Moreover, this unique equilibrium may not be Markovian, indicating that the PDE approach cannot work in this case. We next study the set value of the game, which roughly speaking is the set of insider's values over all equilibria and thus is by nature unique. We show that, although the bridge type of equilibria in the literature does not satisfy the required integrability for our equilibria, its truncation serves as a desired approximate equilibrium and its value belongs to our set value. Finally, we characterize our set value through a level set of certain standard HJB equation.

math.OC↗