Search arXiv⌕ Search

arXiv subjects

Han Cui

Publications and source records attributed to Han Cui.

At least 19 recordsLinked to original sources
arXiv2026

FAAST: Forward-Only Associative Learning via Closed-Form Fast Weights for Test-Time Supervised Adaptation

Adapting pretrained models typically involves a trade-off between the high training costs of backpropagation and the heavy inference overhead of memory-based or in-context learning. We propose FAAST, a forward-only associative adaptation method that analytically compiles labeled examples into fast weights in a single pass. By eliminating memory or context dependence, FAAST achieves constant-time inference and decouples task adaptation from pretrained representation. Across image classification and language modeling benchmarks, FAAST matches or exceeds backprop-based adaptation while reducing adaptation time by over 90% and is competitive to memory/context-based adaptation while saving memory usage by up to 95%. These results demonstrate FAAST as a highly efficient, scalable solution for supervised task adaptation, particularly for resource-constrained models. We release the code and models at https://github.com/baoguangsheng/faast.

cs.LG↗
arXiv2026

Channel Estimation and LDPC Decoding for Bursty Phase Noise

Time-varying distortions in communication systems can significantly degrade the performance of soft-decision forward error correction. This paper presents a burst-aware (BA) low-density parity-check (LDPC) decoding scheme for channels affected by bursty phase noise. By applying differential coding to a Wiener process with time-varying innovation variance, bursty differential phase noise is obtained. Simulation results demonstrate that, compared to conventional decoding, the BA scheme achieves gains in the signal-to-noise ratio of up to $0.7$~dB at a bit error rate (BER) of $4\cdot10^{-3}$ and more than $1$~dB at a packet error rate (PER) of $1\cdot10^{-2}$. Furthermore, by iterating between channel estimation and \ac{ldpc} decoding, forming the proposed iterative burst-aware (IBA) decoding scheme, the gains increase to $1.4$~dB and more than $3$~dB, respectively. More importantly, the IBA scheme significantly improves robustness to bursty phase noise. Compared with the conventional scheme, the IBA scheme can reduce both \ac{ber} and \ac{per} by up to two orders of magnitude under severe bursty phase noise.

eess.SP↗
arXiv2026

Minimizing Mismatch Risk: A Prototype-Based Routing Framework for Zero-shot LLM-generated Text Detection

Zero-shot methods detect LLM-generated text by computing statistical signatures using a surrogate model. Existing approaches typically employ a fixed surrogate for all inputs regardless of the unknown source. We systematically examine this design and find that detection performance varies substantially depending on surrogate-source alignment. We observe that while no single surrogate achieves optimal performance universally, a well-matched surrogate typically exists within a diverse pool for any given input. This finding transforms robust detection into a routing problem: selecting the most appropriate surrogate for each input. We propose DetectRouter, a prototype-based framework that learns text-detector affinity through two-stage training. The first stage constructs discriminative prototypes from white-box models; the second generalizes to black-box sources by aligning geometric distances with observed detection scores. Experiments on EvoBench and MAGE benchmarks demonstrate consistent improvements across multiple detection criteria and model families.

cs.CL↗
arXiv2026

When AI Settles Down: Late-Stage Stability as a Signature of AI-Generated Text Detection

Zero-shot detection methods for AI-generated text typically aggregate token-level statistics across entire sequences, overlooking the temporal dynamics inherent to autoregressive generation. We analyze over 120k text samples and reveal Late-Stage Volatility Decay: AI-generated text exhibits rapidly stabilizing log probability fluctuations as generation progresses, while human writing maintains higher variability throughout. This divergence peaks in the second half of sequences, where AI-generated text shows 24--32\% lower volatility. Based on this finding, we propose two simple features: Derivative Dispersion and Local Volatility, which computed exclusively from late-stage statistics. Without perturbation sampling or additional model access, our method achieves state-of-the-art performance on EvoBench and MAGE benchmarks and demonstrates strong complementarity with existing global methods.

cs.CL↗
arXiv2025

Lead-Free Europium Halide Perovskite Nanoplatelets

Metal halide perovskites possess desirable optical, material, and electrical properties which have had substantial impact on next-generation optoelectronics. However, given the toxicity of lead, alternative lead-free perovskite semiconductors are needed. By fully replacing lead with rare-earth elements, one can simultaneously address toxicity concerns and achieve comparable optoelectronic performance. Here, we demonstrate the synthesis of two-dimensional europium halide perovskite nanoplatelets governed by the formula $\mathrm{L_{2}EuX_{4}}$ where L is an organic ligand and X is a halide anion. The structure, morphology, and composition of the nanoplatelets are confirmed by XRD, AFM, and XPS. Deep blue-emitting $\mathrm{PEA_{2}EuBr_{4}}$ perovskite nanoplatelets are synthesized in both the solution- and solid-states with photoluminescence emission centered at 446 nm and CIE color coordinates of (0.1515, 0.0327) and (0.1515, 0.0342), respectively. Additionally, near-ultraviolet $\mathrm{PEA_{2}EuCl_{4}}$ perovskite nanoplatelets are synthesized in both the solution- and solid-states with photoluminescence emission centered at 400 nm and 401 nm, respectively. Overall, europium halide perovskite nanoplatelets offer a lead-free alternative for deep blue, violet, and near-ultraviolet light emission $\unicode{x2013}$ charting new pathways for optoelectronics in this energy regime.

cond-mat.mtrl-sci↗
arXiv2025

Deep Literature Survey Automation with an Iterative Workflow

Automatic literature survey generation has attracted increasing attention, yet most existing systems follow a one-shot paradigm, where a large set of papers is retrieved at once and a static outline is generated before drafting. This design often leads to noisy retrieval, fragmented structures, and context overload, ultimately limiting survey quality. Inspired by the iterative reading process of human researchers, we propose \ours, a framework based on recurrent outline generation, in which a planning agent incrementally retrieves, reads, and updates the outline to ensure both exploration and coherence. To provide faithful paper-level grounding, we design paper cards that distill each paper into its contributions, methods, and findings, and introduce a review-and-refine loop with visualization enhancement to improve textual flow and integrate multimodal elements such as figures and tables. Experiments on both established and emerging topics show that \ours\ substantially outperforms state-of-the-art baselines in content coverage, structural coherence, and citation quality, while producing more accessible and better-organized surveys. To provide a more reliable assessment of such improvements, we further introduce Survey-Arena, a pairwise benchmark that complements absolute scoring and more clearly positions machine-generated surveys relative to human-written ones. The code is available at https://github.com/HancCui/IterSurvey\_Autosurveyv2.

cs.CL↗
arXiv2025

The Global Well-posedness of the Euler-Poisson System for Ions in 2D

This paper aims to establish the global well-posedness of the Euler-Poisson system for ions in 2D. The difficulties arising from time resonance at low frequencies and slow decay will be overcome by applying the method developed for the gravity-capillary water waves and Euler-Maxwell systems.

math.AP↗
arXiv2025

A Black-Box Debiasing Framework for Conditional Sampling

Conditional sampling is a fundamental task in Bayesian statistics and generative modeling. Consider the problem of sampling from the posterior distribution $P_{X|Y=y^*}$ for some observation $y^*$, where the likelihood $P_{Y|X}$ is known, and we are given $n$ i.i.d. samples $D=\{X_i\}_{i=1}^n$ drawn from an unknown prior distribution $\pi_X$. Suppose that $f(\hat{\pi}_{X^n})$ is the distribution of a posterior sample generated by an algorithm (e.g. a conditional generative model or the Bayes rule) when $\hat{\pi}_{X^n}$ is the empirical distribution of the training data. Although averaging over the randomness of the training data $D$, we have $\mathbb{E}_D\left(\hat{\pi}_{X^n}\right)= \pi_X$, we do not have $\mathbb{E}_D\left\{f(\hat{\pi}_{X^n})\right\}= f(\pi_X)$ due to the nonlinearity of $f$, leading to a bias. In this paper we propose a black-box debiasing scheme that improves the accuracy of such a naive plug-in approach. For any integer $k$ and under boundedness of the likelihood and smoothness of $f$, we generate samples $\hat{X}^{(1)},\dots,\hat{X}^{(k)}$ and weights $w_1,\dots,w_k$ such that $\sum_{i=1}^kw_iP_{\hat{X}^{(i)}}$ is a $k$-th order approximation of $f(\pi_X)$, where the generation process treats $f$ as a black-box. Our generation process achieves higher accuracy when averaged over the randomness of the training data, without degrading the variance, which can be interpreted as improving memorization without compromising generalization in generative models.

stat.ME↗
arXiv2025

Robust Sensitivity Analysis via Augmented Percentile Bootstrap under Simultaneous Violations of Unconfoundedness and Overlap

The identification of causal effects in observational studies typically relies on two standard assumptions: unconfoundedness and overlap. However, both assumptions are often questionable in practice: unconfoundedness is inherently untestable, and overlap may fail in the presence of extreme unmeasured confounding. While various approaches have been developed to address unmeasured confounding and extreme propensity scores separately, few methods accommodate simultaneous violations of both assumptions. In this paper, we propose a sensitivity analysis framework that relaxes both unconfoundedness and overlap, building upon the marginal sensitivity model. Specifically, we allow the bound on unmeasured confounding to hold for only a subset of the population, thereby accommodating heterogeneity in confounding and allowing treatment probabilities to be zero or one. Moreover, unlike prior work, our approach does not require bounded outcomes and focuses on overlap-weighted average treatment effects, which are both practically meaningful and robust to non-overlap. We develop computationally efficient methods to obtain worst-case bounds via linear programming, and introduce a novel augmented percentile bootstrap procedure for statistical inference. This bootstrap method handles parameters defined through over-identified estimating equations involving unobserved variables and may be of independent interest. Our work provides a unified and flexible framework for sensitivity analysis under violations of both unconfoundedness and overlap.

stat.ME↗
arXiv2025

Global well-posedness of the Cauchy problem for the modified Whitham equations

This paper aims to show global existence and modified scattering for the solutions of the Cauchy problem to the modified Whitham equations for small, smooth and localized initial data. The main difficulties come from slow decay and non-homogeneity of the Fourier multiplier $(\sqrt{\tanh \xi/\xi})\xi$, which will be overcome by introducing an interaction multiplier theorem and estimating the weighted norms in the frequency space. When estimating the weighted norms, due to loss of derivatives, the energy estimate will be performed in the frequency space, and the absence of time resonance will be effectively utilized by extracting some good terms arising from integration by parts in time before the energy estimate.

math.AP↗
arXiv2025

Direct Value Optimization: Improving Chain-of-Thought Reasoning in LLMs with Refined Values

We introduce Direct Value Optimization (DVO), an innovative reinforcement learning framework for enhancing large language models in complex reasoning tasks. Unlike traditional methods relying on preference labels, DVO utilizes value signals at individual reasoning steps, optimizing models via a mean squared error loss. The key benefit of DVO lies in its fine-grained supervision, circumventing the need for labor-intensive human annotations. Target values within the DVO are estimated using either Monte Carlo Tree Search or an outcome value model. Our empirical analysis on both mathematical and commonsense reasoning tasks shows that DVO consistently outperforms existing offline preference optimization techniques, even with fewer training steps. These findings underscore the importance of value signals in advancing reasoning capabilities and highlight DVO as a superior methodology under scenarios lacking explicit human preference information.

cs.CL↗
arXiv2024

An Empirical Study of False Negatives and Positives of Static Code Analyzers From the Perspective of Historical Issues

Static code analyzers are widely used to help find program flaws. However, in practice the effectiveness and usability of such analyzers is affected by the problems of false negatives (FNs) and false positives (FPs). This paper aims to investigate the FNs and FPs of such analyzers from a new perspective, i.e., examining the historical issues of FNs and FPs of these analyzers reported by the maintainers, users and researchers in their issue repositories -- each of these issues manifested as a FN or FP of these analyzers in the history and has already been confirmed and fixed by the analyzers' developers. To this end, we conduct the first systematic study on a broad range of 350 historical issues of FNs/FPs from three popular static code analyzers (i.e., PMD, SpotBugs, and SonarQube). All these issues have been confirmed and fixed by the developers. We investigated these issues' root causes and the characteristics of the corresponding issue-triggering programs. It reveals several new interesting findings and implications on mitigating FNs and FPs. Furthermore, guided by some findings of our study, we designed a metamorphic testing strategy to find FNs and FPs. This strategy successfully found 14 new issues of FNs/FPs, 11 of which have been confirmed and 9 have already been fixed by the developers. Our further manual investigation of the studied analyzers revealed one rule specification issue and additional four FNs/FPs due to the weaknesses of the implemented static analysis. We have made all the artifacts (datasets and tools) publicly available at https://zenodo.org/doi/10.5281/zenodo.11525129.

cs.SE↗
arXiv2024

CBCT scatter correction with dual-layer flat-panel detector

Background: Recently, the popularity of dual-layer flat-panel detector (DL-FPD) based dual-energy cone-beam CT (DE-CBCT) imaging has been increasing. However, the image quality of DE-CBCT remains constrained by the Compton scattered X-ray photons. Purpose: The objective of this study is to develop an energy-modulated scatter correction method for DL-FPD based CBCT imaging. Methods: In DL-FPD, a certain portion of the X-ray photons (mainly low-energy primary and scattered photons) passing through the object are captured by the top detector layer, while the remaining X-ray photons (mainly high-energy primary and scattered photons) are collected by the bottom detector layer. Based on the two set of distinct low-energy and high-energy measurements, a linear signal model was approximated for the dual-energy primary and scattered signals on DL-FPD. The distributions of X-ray scatters were quickly estimated using this signal model. Monte Carlo (MC) simulation of a water phantom was conducted to verify this newly developed scatter estimation method. Moreover, physical experiments of water phantom, head phantom, and abdominal phantom were carried out to validate the real performance of this proposed scatter correction method. Results: The MC results showed that the e-Grid method was able to generate scatter distributions close to the ground truth. Moreover, the physical experiments demonstrated that the e-Grid method can greatly reduce the shading artifacts in both low-energy and high-energy CBCT images acquired from DL-FPD. On average, the image non-uniformity (NU) was reduced by over 77% in the low-energy CBCT image and by over 66% in the high-energy CBCT image. A a consequence, the accuracy of the decomposed multi-material bases was substantially improved.

physics.med-ph↗
arXiv2024

Sampling from the Random Linear Model via Stochastic Localization Up to the AMP Threshold

Recently, Approximate Message Passing (AMP) has been integrated with stochastic localization (diffusion model) by providing a computationally efficient estimator of the posterior mean. Existing (rigorous) analysis typically proves the success of sampling for sufficiently small noise, but determining the exact threshold involves several challenges. In this paper, we focus on sampling from the posterior in the linear inverse problem, with an i.i.d. random design matrix, and show that the threshold for sampling coincides with that of posterior mean estimation. We give a proof for the convergence in smoothed KL divergence whenever the noise variance $\Delta$ is below $\Delta_{\rm AMP}$, which is the computation threshold for mean estimation introduced in (Barbier et al., 2020). We also show convergence in the Wasserstein distance under the same threshold assuming a dimension-free bound on the operator norm of the posterior covariance matrix, a condition strongly suggested by recent breakthroughs on operator norm bounds in similar replica symmetric systems. A key observation in our analysis is that phase transition does not occur along the sampling and interpolation paths assuming $\Delta<\Delta_{\rm AMP}$.

math.ST↗
arXiv2024

Reference-Free Image Quality Metric for Degradation and Reconstruction Artifacts

Image Quality Assessment (IQA) is essential in various Computer Vision tasks such as image deblurring and super-resolution. However, most IQA methods require reference images, which are not always available. While there are some reference-free IQA metrics, they have limitations in simulating human perception and discerning subtle image quality variations. We hypothesize that the JPEG quality factor is representatives of image quality measurement, and a well-trained neural network can learn to accurately evaluate image quality without requiring a clean reference, as it can recognize image degradation artifacts based on prior knowledge. Thus, we developed a reference-free quality evaluation network, dubbed "Quality Factor (QF) Predictor", which does not require any reference. Our QF Predictor is a lightweight, fully convolutional network comprising seven layers. The model is trained in a self-supervised manner: it receives JPEG compressed image patch with a random QF as input, is trained to accurately predict the corresponding QF. We demonstrate the versatility of the model by applying it to various tasks. First, our QF Predictor can generalize to measure the severity of various image artifacts, such as Gaussian Blur and Gaussian noise. Second, we show that the QF Predictor can be trained to predict the undersampling rate of images reconstructed from Magnetic Resonance Imaging (MRI) data.

eess.IV↗
arXiv2024

Exploring Hybrid Question Answering via Program-based Prompting

Question answering over heterogeneous data requires reasoning over diverse sources of data, which is challenging due to the large scale of information and organic coupling of heterogeneous data. Various approaches have been proposed to address these challenges. One approach involves training specialized retrievers to select relevant information, thereby reducing the input length. Another approach is to transform diverse modalities of data into a single modality, simplifying the task difficulty and enabling more straightforward processing. In this paper, we propose HProPro, a novel program-based prompting framework for the hybrid question answering task. HProPro follows the code generation and execution paradigm. In addition, HProPro integrates various functions to tackle the hybrid reasoning scenario. Specifically, HProPro contains function declaration and function implementation to perform hybrid information-seeking over data from various sources and modalities, which enables reasoning over such data without training specialized retrievers or performing modal transformations. Experimental results on two typical hybrid question answering benchmarks HybridQA and MultiModalQA demonstrate the effectiveness of HProPro: it surpasses all baseline systems and achieves the best performances in the few-shot settings on both datasets.

cs.CL↗
arXiv2023

MiliPoint: A Point Cloud Dataset for mmWave Radar

Millimetre-wave (mmWave) radar has emerged as an attractive and cost-effective alternative for human activity sensing compared to traditional camera-based systems. mmWave radars are also non-intrusive, providing better protection for user privacy. However, as a Radio Frequency (RF) based technology, mmWave radars rely on capturing reflected signals from objects, making them more prone to noise compared to cameras. This raises an intriguing question for the deep learning community: Can we develop more effective point set-based deep learning methods for such attractive sensors? To answer this question, our work, termed MiliPoint, delves into this idea by providing a large-scale, open dataset for the community to explore how mmWave radars can be utilised for human activity recognition. Moreover, MiliPoint stands out as it is larger in size than existing datasets, has more diverse human actions represented, and encompasses all three key tasks in human activity recognition. We have also established a range of point-based deep neural networks such as DGCNN, PointNet++ and PointTransformer, on MiliPoint, which can serve to set the ground baseline for further development.

cs.LG↗
arXiv2023

Super resolution dual-layer CBCT imaging with model-guided deep learning

Objective: This study aims at investigating a novel super resolution CBCT imaging technique with the dual-layer flat panel detector (DL-FPD). Approach: In DL-FPD based CBCT imaging, the low-energy and high-energy projections acquired from the top and bottom detector layers contain intrinsically mismatched spatial information, from which super resolution CBCT images can be generated. To explain, a simple mathematical model is established according to the signal formation procedure in DL-FPD. Next, a dedicated recurrent neural network (RNN), named as suRi-Net, is designed by referring to the above imaging model to retrieve the high resolution dual-energy information. Different phantom experiments are conducted to validate the performance of this newly developed super resolution CBCT imaging method. Main Results: Results show that the proposed suRi-Net can retrieve high spatial resolution information accurately from the low-energy and high-energy projections having lower spatial resolution. Quantitatively, the spatial resolution of the reconstructed CBCT images of the top and bottom detector layers is increased by about 45% and 54%, respectively. Significance: In future, suRi-Net provides a new approach to achieve high spatial resolution dual-energy imaging in DL-FPD based CBCT systems.

physics.med-ph↗