arXiv subjects

Ke Jia

Publications and source records attributed to Ke Jia.

11 recordsLinked to original sources

arXiv2026

Amplitude-Only FFN Intervention for Tool-Structured LLM Inference Method: Gated Evaluation Protocol, and Cross-Model Empirical Results

Large language models increasingly operate as tool-using agents, where small format, argument, or function-call errors can invalidate otherwise plausible responses. We study inference-time feed-forward network (FFN) intervention as a way to improve structured outputs without retraining model weights. An earlier project-specific approach, Orthogonal Residual Projection (ORP), exposed sensitive SwiGLU FFN sites and non-monotonic energy effects, but its direction-changing operation produced more regressions than repairs in a key diagnostic. We therefore propose Amplitude Gating (AG), which preserves pretrained FFN weight directions and modulates activation magnitudes during decoding. AG separates candidate generation, ranking, and a prospective acceptance/fallback decision. We also introduce Per-Sample Fix-Harm Evaluation (PFHE), a paired reporting protocol that complements native task metrics with fixes, harms, preserved-correct cases, and preserved-wrong cases. On the only cross-position union that passes source-alignment audit, an exploratory offline mixed selector raises the descriptive heterogeneous-scorer Qwen3.5-9B tool-route micro-average from 38.66% to 42.92% (+4.27 percentage points); two Hermes function-call endpoints improve by +7.64 and +7.62 points. The same-output PFHE-format view records 48 fixes, 26 harms, 294 preserved-correct cases, and 2,188 preserved-wrong cases over 2,556 units, with positive paired bootstrap intervals for native and strict effects. Protocol-separated Qwen3-8B and Qwen2.5-7B analyses retain oracle headroom but no positive train-selected fixed tool route. A grouped five-fold RF diagnostic suggests weak nonlinear ranking signal but forces intervention, lacks baseline fallback and paired uncertainty, and is not deployment evidence. The results support model- and task-specific selection with strict fallback, not a universal AG switch.

cs.CL↗

arXiv2025

PrAda-GAN: A Private Adaptive Generative Adversarial Network with Bayes Network Structure

We revisit the problem of generating synthetic data under differential privacy. To address the core limitations of marginal-based methods, we propose the Private Adaptive Generative Adversarial Network with Bayes Network Structure (PrAda-GAN), which integrates the strengths of both GAN-based and marginal-based approaches. Our method adopts a sequential generator architecture to capture complex dependencies among variables, while adaptively regularizing the learned structure to promote sparsity in the underlying Bayes network. Theoretically, we establish diminishing bounds on the parameter distance, variable selection error, and Wasserstein distance. Our analysis shows that leveraging dependency sparsity leads to significant improvements in convergence rates. Empirically, experiments on both synthetic and real-world datasets demonstrate that PrAda-GAN outperforms existing tabular data synthesis methods in terms of the privacy-utility trade-off.

stat.ML↗

arXiv2025

An Efficient Deep Template Matching and In-Plane Pose Estimation Method via Template-Aware Dynamic Convolution

In industrial inspection and component alignment tasks, template matching requires efficient estimation of a target's position and geometric state (rotation and scaling) under complex backgrounds to support precise downstream operations. Traditional methods rely on exhaustive enumeration of angles and scales, leading to low efficiency under compound transformations. Meanwhile, most deep learning-based approaches only estimate similarity scores without explicitly modeling geometric pose, making them inadequate for real-world deployment. To overcome these limitations, we propose a lightweight end-to-end framework that reformulates template matching as joint localization and geometric regression, outputting the center coordinates, rotation angle, and independent horizontal and vertical scales. A Template-Aware Dynamic Convolution Module (TDCM) dynamically injects template features at inference to guide generalizable matching. The compact network integrates depthwise separable convolutions and pixel shuffle for efficient matching. To enable geometric-annotation-free training, we introduce a rotation-shear-based augmentation strategy with structure-aware pseudo labels. A lightweight refinement module further improves angle and scale precision via local optimization. Experiments show our 3.07M model achieves high precision and 14ms inference under compound transformations. It also demonstrates strong robustness in small-template and multi-object scenarios, making it highly suitable for deployment in real-time industrial applications. The code is available at:https://github.com/ZhouJ6610/PoseMatch-TDCM.

cs.CV↗

arXiv2024

Better Locally Private Sparse Estimation Given Multiple Samples Per User

Previous studies yielded discouraging results for item-level locally differentially private linear regression with $s^*$-sparsity assumption, where the minimax rate for $nm$ samples is $\mathcal{O}(s^{*}d / nm\varepsilon^2)$. This can be challenging for high-dimensional data, where the dimension $d$ is extremely large. In this work, we investigate user-level locally differentially private sparse linear regression. We show that with $n$ users each contributing $m$ samples, the linear dependency of dimension $d$ can be eliminated, yielding an error upper bound of $\mathcal{O}(s^{*2} / nm\varepsilon^2)$. We propose a framework that first selects candidate variables and then conducts estimation in the narrowed low-dimensional space, which is extendable to general sparse estimation problems with tight error bounds. Experiments on both synthetic and real datasets demonstrate the superiority of the proposed methods. Both the theoretical and empirical results suggest that, with the same number of samples, locally private sparse estimation is better conducted when multiple samples per user are available.

stat.ML↗

arXiv2024

Locally Private Estimation with Public Features

We initiate the study of locally differentially private (LDP) learning with public features. We define semi-feature LDP, where some features are publicly available while the remaining ones, along with the label, require protection under local differential privacy. Under semi-feature LDP, we demonstrate that the mini-max convergence rate for non-parametric regression is significantly reduced compared to that of classical LDP. Then we propose HistOfTree, an estimator that fully leverages the information contained in both public and private features. Theoretically, HistOfTree reaches the mini-max optimal convergence rate. Empirically, HistOfTree achieves superior performance on both synthetic and real data. We also explore scenarios where users have the flexibility to select features for protection manually. In such cases, we propose an estimator and a data-driven parameter tuning strategy, leading to analogous theoretical and empirical results.

stat.ML↗

arXiv2024

Discovery of a Magnetic Topological Semimetal Eu$_3$In$_2$As$_4$ with a Single Pair of Weyl Points

Magnetic Weyl semimetal (MWS) is a unique topological state with open surface Fermi arc states and other exotic transport phenomena. However, most reported MWSs show multiple pairs of Weyl points and complicated Fermi surfaces, which increases the difficulty of the investigation into the intrinsic chiral transport property. In this wor, we successfully synthesized a soft magnetic Weyl semimetal Eu$_3$In$_2$As$_4$ with a single pair of Weyl points under magnetic fields. The Shubnikov de Haas (SdH) oscillation with a single frequency, as well as a linear hall resistance with the same carrier density, is observed up to 50 Tesla, indicating a single pair of Weyl points around the Fermi level with a massless fermion ($m^* = 0.121 m_0$, $\pi$ Berry phase). Such a single pair of Weyl points is further confirmed by the density functional theory calculations. The magnetic ordering and band topology can be easily tuned by the external magnetic field. The field-induced MWS Eu$_3$In$_2$As$_4$ with a single pair of Weyl points is a good platform to detect chiral transport properties, including possible quantum anomalous Hall effect.

cond-mat.mes-hall↗

arXiv2023

Tunneling spectroscopic signatures of charge doping and associated Mott transition in $α$-RuCl${_3}$ in proximity to graphite

The layered Mott insulator $α$-RuCl${_3}$ has been extensively studied as a potential Kitaev quantum spin liquid candidate. Here, by constructing heterostructures with graphite, we employed electron tunneling measurements on few-layer $α$-RuCl${_3}$ using a scanning tunneling microscopy/spectroscopy. Characteristic tunneling spectra were detected on $α$-RuCl${_3}$ layers in proximity to graphite. In the single-layer $α$-RuCl${_3}$ in direct contact with graphite, distinct states in the Mott-gap regime were observed. The in-gap states are demonstrated to be closely related to the electron orbitals in $α$-RuCl${_3}$ and graphite, and to be sensitive to interfacial coupling, where a hybridization at the heterointerface is hypothesized. The in-gap states are also thought of as a charge reservoir for weakly doping the $α$-RuCl${_3}$ upper-layers. It demonstrated that the weak doping effect causes a considerable decrease in the Mott-gap within the upper-layers, suggesting that an unconventional Mott-transition is occurring in these layers. The results show that the heterostructure comprised of $α$-RuCl${_3}$ and graphite is a good platform for investigating the doping physics in $α$-RuCl${_3}$. Therefore, tunneling into such a doped system is a useful probe for studying otherwise insulating spin-liquid candidates.

cond-mat.str-el↗

arXiv2023

Inverse-current quantum electro-oscillations in a charge-density wave insulator

Quantum magneto-oscillations have long been a vital subject in condensed matter physics, with ubiquitous quantum phenomena and diverse underlying physical mechanisms. Here, we demonstrate the intrinsic and reproducible DC-current-driven quantum electro-oscillations with a periodicity in the inverse of the current (1/I), in quasi-one-dimensional charge-density-wave (CDW) insulators (TaSe$_4$)$_2$I and TaS$_3$ nanowires. Such oscillations manifest in the nearly infinite Fr\"ohlich conductivity region where the undamped CDW flow forms in a finite electric current, and finally disappear after the oscillation index n reaches 1. A systematic investigation on the effect of temperature and magnetic field establishes that the observed electro-oscillations are a coherent quantum phenomenon. We discuss the possibilities of the physical mechanisms, including the formation of sliding-driven inherent Floquet sidebands. Our results introduce a new member in the family of quantum oscillations, and shed light on plausible avenues to explore novel physics and potential applications of coherent density-wave condensates.

cond-mat.mes-hall↗

arXiv2022

Neutron-diffraction and linear {Grüneisen} parameter studies of magnetism in NdFe$_2$Ga$_8$

We study the magnetism in NdFe$_2$Ga$_8$ by the neutron-diffraction and temperature-modulated linear {Grüneisen} parameter measurements. Previous thermodynamical measurements have demonstrated that there are two magnetic transitions at 10 and 14.5 K, respectively. Neutron-diffraction measurements confirm that the lower one is an antiferromagnetic (AFM) transition with a commensurate magnetic structure. Both the commensurate and the incommensurate (IC) magnetic peaks are found below the higher transition but their intensities only gradually increase with decreasing temperature. Below 10 K, the commensurate peak intensity increases quickly with decreasing temperature, signaling the AFM transition, while the IC peak intensity disappears below 5 K. The linear {Grüneisen} parameter along the $c$ axis, $Γ_c$, shows a hysteresis behavior that is different from the hysteresis behavior for the magnetization $M$. We give a discussion of the origin of the magnetism in NdFe$_2$Ga$_8$.

cond-mat.str-el↗

arXiv2022

Antiferromagnetic Excitonic Insulator State in Sr3Ir2O7

Excitonic insulators are usually considered to form via the condensation of a soft charge mode of bound electron-hole pairs. This, however, presumes that the soft exciton is of spin-singlet character. Early theoretical considerations have also predicted a very distinct scenario, in which the condensation of magnetic excitons results in an antiferromagnetic excitonic insulator state. Here we report resonant inelastic x-ray scattering (RIXS) measurements of Sr3Ir2O7. By isolating the longitudinal component of the spectra, we identify a magnetic mode that is well-defined at the magnetic and structural Brillouin zone centers, but which merges with the electronic continuum in between these high-symmetry points and which decays upon heating concurrent with a decrease in the material's resistivity. We show that a bilayer Hubbard model, in which electron-hole pairs are bound by exchange interactions, consistently explains all the electronic and magnetic properties of Sr3Ir2O7 indicating that this material is a realization of the long-predicted antiferromagnetic excitonic insulators phase.

cond-mat.str-el↗

arXiv2019

Road Segmentation with Image-LiDAR Data Fusion

Robust road segmentation is a key challenge in self-driving research. Though many image-based methods have been studied and high performances in dataset evaluations have been reported, developing robust and reliable road segmentation is still a major challenge. Data fusion across different sensors to improve the performance of road segmentation is widely considered an important and irreplaceable solution. In this paper, we propose a novel structure to fuse image and LiDAR point cloud in an end-to-end semantic segmentation network, in which the fusion is performed at decoder stage instead of at, more commonly, encoder stage. During fusion, we improve the multi-scale LiDAR map generation to increase the precision of the multi-scale LiDAR map by introducing pyramid projection method. Additionally, we adapted the multi-path refinement network with our fusion strategy and improve the road prediction compared with transpose convolution with skip layers. Our approach has been tested on KITTI ROAD dataset and has competitive performance.

cs.CV↗