arXiv subjects

Ruihan Hu

Publications and source records attributed to Ruihan Hu.

7 recordsLinked to original sources

arXiv2026

FluxBin: Flexible LUT-based Ultra-low-bit LLM Inference by Algorithm-Kernel Synergy

While binary quantization theoretically promises extreme compression and acceleration for Large Language Models (LLMs), existing research often overlooks the necessity of specialized hardware kernels, thus failing to unleash the full acceleration potential due to persistent reliance on expensive floating-point arithmetic or runtime dequantization overheads. To bridge this gap, we propose FluxBin (\textbf{F}lexible \textbf{L}UT-based \textbf{U}ltra-low-bit e\textbf{X}ecution with \textbf{Bin}ary bases), an algorithm-kernel co-design that synergizes post-training quantization with a highly optimized CUDA kernel. Algorithmically, we introduce Decoupled Row-Column Binary Decomposition to enhance representational capacity while maintaining hardware efficiency, complemented by a Hessian-guided saliency-aware hybrid bases that preserve critical information. At the kernel level, we implement a Lookup Table Building Approach with Scale Fusion to reduce floating-point arithmetic, featuring a Virtual Columnar Mapping that transforms irregular, sparse, and salient matrices into dense execution. Extensive evaluations demonstrate FluxBin achieves up to $5.92\times$ speedup and $10.19\times$ energy savings across diverse model architectures, delivering comparable accuracy to heavily fine-tuned methods. This effectively enables the deployment of 70B-scale models on one single A100 GPU with a $4\times$ memory reduction. Code is available at https://github.com/nicyyyy/FluxBin.

cs.LG↗

arXiv2026

When Reasoning Leaks Membership: Membership Inference Attack on Black-box Large Reasoning Models

Large Reasoning Models (LRMs) have rapidly gained prominence for their strong performance in solving complex tasks. Many modern black-box LRMs expose the intermediate reasoning traces through APIs to improve transparency (e.g., Gemini-2.5 and Claude-sonnet). Despite their benefits, we find that these traces can leak membership signals, creating a new privacy threat even without access to token logits used in prior attacks. In this work, we initiate the first systematic exploration of Membership Inference Attacks (MIAs) on black-box LRMs. Our preliminary analysis shows that LRMs produce confident, recall-like reasoning traces on familiar training member samples but more hesitant, inference-like reasoning traces on non-members. The representations of these traces are continuously distributed in the semantic latent space, spanning from familiar to unfamiliar samples. Building on this observation, we propose BlackSpectrum, the first membership inference attack framework targeting the black-box LRMs. The key idea is to construct a recall-inference axis in the semantic latent space, based on representations derived from the exposed traces. By locating where a query sample falls along this axis, the attacker can obtain a membership score and predict how likely it is to be a member of the training data. Additionally, to address the limitations of outdated datasets unsuited to modern LRMs, we provide two new datasets to support future research, arXivReasoning and BookReasoning. Empirically, exposing reasoning traces significantly increases the vulnerability of LRMs to membership inference attacks, leading to large gains in attack performance. Our findings highlight the need for LRM companies to balance transparency in intermediate reasoning traces with privacy preservation.

cs.CR↗

arXiv2025

From static to adaptive: immune memory-based jailbreak detection for large language models

Large Language Models (LLMs) serve as the backbone of modern AI systems, yet they remain susceptible to adversarial jailbreak attacks. Consequently, robust detection of such malicious inputs is paramount for ensuring model safety. Traditional detection methods typically rely on external models trained on fixed, large-scale datasets, which often incur significant computational overhead. While recent methods shift toward leveraging internal safety signals of models to enable more lightweight and efficient detection. However, these methods remain inherently static and struggle to adapt to the evolving nature of jailbreak attacks. Drawing inspiration from the biological immune mechanism, we introduce the Immune Memory Adaptive Guard (IMAG) framework. By distilling and encoding safety patterns into a persistent, evolvable memory bank, IMAG enables adaptive generalization to emerging threats. Specifically, the framework orchestrates three synergistic components: Immune Detection, which employs retrieval for high-efficiency interception of known jailbreak attacks; Active Immunity, which performs proactive behavioral simulation to resolve ambiguous unknown queries; Memory Updating, which integrates validated attack patterns back into the memory bank. This closed-loop architecture transitions LLM defense from rigid filtering to autonomous adaptive mitigation. Extensive evaluations across five representative open-source LLMs demonstrate that our method surpasses state-of-the-art (SOTA) baselines, achieving a superior average detection accuracy of 94\% across diverse and complex attack types.

cs.CR↗

arXiv2025

Automated Detection of Pre-training Text in Black-box LLMs

Detecting whether a given text is a member of the pre-training data of Large Language Models (LLMs) is crucial for ensuring data privacy and copyright protection. Most existing methods rely on the LLM's hidden information (e.g., model parameters or token probabilities), making them ineffective in the black-box setting, where only input and output texts are accessible. Although some methods have been proposed for the black-box setting, they rely on massive manual efforts such as designing complicated questions or instructions. To address these issues, we propose VeilProbe, the first framework for automatically detecting LLMs' pre-training texts in a black-box setting without human intervention. VeilProbe utilizes a sequence-to-sequence mapping model to infer the latent mapping feature between the input text and the corresponding output suffix generated by the LLM. Then it performs the key token perturbations to obtain more distinguishable membership features. Additionally, considering real-world scenarios where the ground-truth training text samples are limited, a prototype-based membership classifier is introduced to alleviate the overfitting issue. Extensive evaluations on three widely used datasets demonstrate that our framework is effective and superior in the black-box setting.

cs.CL↗

arXiv2020

A Multilayer Neural Network Merging Image Preprocessing and Pattern Recognition by Integrating Diffusion and Drift Memristors

With the development of research on novel memristor model and device, neural networks by integrating various memristor models have become a hot research topic recently. However, state-of-the-art works still build such neural networks using drift memristor only. Furthermore, some other related works are only applied to a few individual applications including pattern recognition and edge detection. In this paper, a novel kind of multilayer neural network is proposed, in which diffusion and drift memristor models are applied to construct a system merging image preprocessing and pattern recognition. Specifically, the entire network consists of two diffusion memristive cellular layers for image preprocessing and one drift memristive feedforward layer for pattern recognition. Experimental results show that good recognition accuracy of noisy MNIST is obtained due to the fusion of image preprocessing and pattern recognition. Moreover, owing to high-efficiency in-memory computing and brief spiking encoding methods, high processing speed, high throughput, and few hardware resources of the entire network are achieved.

cs.ET↗

arXiv2020

Fully Memristive Spiking-Neuron Learning Framework and its Applications on Pattern Recognition and Edge Detection

Fully memristive spiking-neuron learning framework, which uses drift and diffusion memristor models as axon and dendrite respectively, becomes a hot topic recently with the development of memristor devices. Normally, some other devices like resistor or capacitor are still necessary on recent works of fully memristive learning framework. However, theoretically, one neuron needs axon and dendrite only, which makes technique process simpler and learning framework more similar to biologic brain. In this paper, a fully memristive spiking-neuron learning framework is introduced, in which a neuron structure is just built of one drift and one diffusion memristive models. To verify it merits, a feedforward neural network for pattern recognition and a cellular neural network for edge detection are designed. Experiment results show that compared to other memristive neural networks, our framework's the processing speed is much faster and the hardware resource is saved in pattern recognition due to its simple structure. Further due to the dynamic filtering function of diffusion memristor model in our learning framework, its peak signal noise ratio (PSNR) is much higher than traditional algorithms in edge detection.

cs.ET↗

arXiv2019

The MBPEP: a deep ensemble pruning algorithm providing high quality uncertainty prediction

Machine learning algorithms have been effectively applied into various real world tasks. However, it is difficult to provide high-quality machine learning solutions to accommodate an unknown distribution of input datasets; this difficulty is called the uncertainty prediction problems. In this paper, a margin-based Pareto deep ensemble pruning (MBPEP) model is proposed. It achieves the high-quality uncertainty estimation with a small value of the prediction interval width (MPIW) and a high confidence of prediction interval coverage probability (PICP) by using deep ensemble networks. In addition to these networks, unique loss functions are proposed, and these functions make the sub-learners available for standard gradient descent learning. Furthermore, the margin criterion fine-tuning-based Pareto pruning method is introduced to optimize the ensembles. Several experiments including predicting uncertainties of classification and regression are conducted to analyze the performance of MBPEP. The experimental results show that MBPEP achieves a small interval width and a low learning error with an optimal number of ensembles. For the real-world problems, MBPEP performs well on input datasets with unknown distributions datasets incomings and improves learning performance on a multi task problem when compared to that of each single model.

cs.LG↗