Searcharxiv⌕ Search

arXiv subjects

Shuhan Yuan

Publications and source records attributed to Shuhan Yuan.

At least 19 recordsLinked to original sources

MIRAGE: Motion Identity Reduction via Autoregressive Generative Encoding for Privacy-Preserving Skeleton-based Motion Data

Skeleton motion identifies people through body proportions and movement style, and skeleton streams are increasingly passed to third-party analytics that run action classifiers trained on raw data. A prior adversarial anonymizer adds a per-frame residual in the raw coordinate space but sees one frame at a time; motion retargeting needs a target skeleton and complete sequences, which rules out online use. We present MIRAGE, a target-free, causal anonymizer that adds an identity-reducing residual to each incoming frame in the raw coordinate space, computed by attention over past frames. One causal, optionally windowed mask governs every attention in a rotary-position encoder-decoder, and key/value caches let the trained function run frame by frame; on all 56,578 NTU 60 sequences, batched and streaming outputs differ by at most 7.2e-6 m. Reported results use an unbounded window, so per-stream state grows with stream length. We evaluate against a passive adversary that applies raw-trained identity models and an adaptive adversary that trains on the anonymizer's output, treating the adaptive adversary as the headline, on NTU RGB+D 60 over five generator seeds, alongside the per-frame residual anonymizer of Moon et al. trained identically. When consumers retrain on the released data, MIRAGE keeps 83.9% action recognition (raw 89.1%; Moon et al. 88.7%) while a retrained identity classifier reaches 43.4% re-identification (RI) over 40 subjects, against 69.4% for Moon et al. and 75.4% on raw data. The strongest adaptive attacker still reaches 49.3% (chance 2.5%), so MIRAGE reduces identity leakage; it does not anonymize. Against raw-trained (frozen) classifiers, MIRAGE lowers RI below the strongest retargeting baseline (10.36% vs. 22.09%), but frozen action recognition falls to 37.04% (vs. 75.82%), so at full residual scale it is not a drop-in replacement for raw input to existing frozen models.

cs.CV↗

CellularS2-Bench: A Staged, Evidence-Grounded Benchmark for Cellular Network Security Reasoning

Cellular networks are critical infrastructure supporting billions of worldwide users and safety- and mission-critical services. Vulnerabilities in cellular networks can therefore cause service disruption, privacy breaches, and broad societal harm, motivating growing efforts to analyze 3GPP specifications that define required device and operator behavior. However, existing benchmarks for this domain share three systematic deficiencies: they rely on LLM-generated entries without expert verification, they do not require faithful interpretation of normative language, reasoning across cross-referenced clauses, and verifiable conclusions grounded in multimodal evidence such as tables and figures, and they do not evaluate security reasoning over end-to-end protocol behavior. To address these gaps, we present CellularS2-Bench, a staged benchmark. CellularS2-Bench covers core control-plane specifications, processing over 1,866 paragraphs, 22,033 sentences, 424 figures, and 248 tables. It comprises 13,851 questions and instances across three progressive stages, of which 10,269 (74.14%) are newly constructed from 3GPP specifications and 43 real-world vulnerabilities. Beyond newly constructed tasks, CellularS2-Bench integrates and corrects 3,582 (25.86%) instances from 7 prior datasets to broaden task coverage. To establish a reproducible baseline on CellularS2-Bench, we introduce CellularS2-ARI, a unified Adapt-Retrieve-Integrate framework for systematic understanding and specification-driven security analysis of 3GPP specifications. Together, they establish an accessible and reproducible foundation for quantifying progress in specification understanding and security reasoning in cellular network security.

cs.CR↗

CellSecInspector: Safeguarding Cellular Networks via Automated Security Analysis on Specifications

The complexity, interdependence, and rapid evolution of 3GPP specifications present fundamental challenges for ensuring the security of modern cellular networks. Manual reviews and existing automated approaches, which often depend on rule-based parsing or small sets of manually crafted security requirements, fail to capture deep semantic dependencies, cross-sentence/clause relationships, and evolving specification behaviors. In this work, we present CellSecInspector, an automated framework for security analysis of 3GPP specifications. CellSecInspector extracts structured state-condition-action (SCA) representations, models mobile network procedures with comprehensive function chains, systematically validates them against 9 foundational security properties under 4 adversarial scenarios, and automatically generates test cases. This end-to-end approach enables the automated discovery of vulnerabilities without relying on manually predefined security requirements or rules. Applying CellSecInspector to the well-studied 5G and 4G NAS and RRC specifications and selected sections of TS 23.501 and TS 24.229, it discovers 43 vulnerabilities, 7 of which are previously unreported. Our findings show that CellSecInspector is a scalable, adaptive, and effective solution to assess 3GPP specifications for safeguarding operational and next-generation cellular networks.

cs.CR↗

MoSign: Challenge-Response Motion-Watermark Authentication for Anonymous Virtual-Reality Users

Social virtual reality (VR) creates a paradox. A user's body motion is a high-entropy biometric: head and hand trajectories alone re-identify users among tens of thousands with over $94\%$ accuracy, so anonymizing the rendered avatar is a practical necessity. Yet a user often still wants to prove their identity to a chosen party from inside that anonymity. We present MoSign, which recasts digital watermarking as a challenge-response authentication protocol on the motion channel. MoSign embeds a time-varying keyed message into the style latent of a motion variational autoencoder via keystream-whitened Gaussian-Shading: watermarked motion is provably indistinguishable from watermark-free motion, since any detector's advantage reduces to breaking a pseudorandom function, so the mark composes with anonymization. The message is a keyed MAC over an epoch counter, a session nonce, and a deployment context, making MoSign replay-resistant and bounding forgery by the verifier's measured false-accept rate times the adversary's online query budget. A key-holding verifier decides with a sequential test. We identify render$\rightarrow$record$\rightarrow$re-estimate ("recapture") as the realistic VR attack surface: a generic pose estimator strips the necessarily subtle watermark, but a recapture-robust keyed reader recovers it (up to $0.96$ codeword accuracy on a projected-2D channel, $0.81$ through a full render-to-video loop), while without the key recovery stays at chance. On HumanML3D, MoSign authenticates every legitimate user at a false-accept rate of $10^{-4}$ on clean and most channels and stays undetectable (detection AUC $0.51$, chance $0.5$); on the BOXRR-23 VR dataset it carries the mark through a real anonymizer at $0.99$ codeword accuracy and adds no de-anonymization side channel.

cs.CR↗

Memorization Is Not Extraction: Tight Differential-Privacy Bounds and Audit Blind Spots

Memorization in large language models is measured through a zoo of definitions whose formal relations are unknown, and differential privacy (DP) is treated as a proxy against all of them at once. We pin down the exact DP constant for the two that carry the practical weight, counterfactual memorization and adaptive extraction, and show that they do not control each other. Under $f$-DP, every adaptive extraction protocol with list budget $m$ succeeds with probability at most $1-f(κ)$ for the oblivious baseline $κ$, and the bound is tight on a dense set of baselines: DP uniformly controls extraction exactly up to a threshold in how well the secret can be guessed a priori. Min-entropy certifies that baseline distribution-free, since $H_\infty\geε\log_2 e+\log_2(m/τ)$ holds extraction below a risk level $τ\le1/2$ under pure $ε$-DP for every prior, and is exact on uniform priors. On the memorization side, $f$-DP caps the counterfactual memorization of any bounded score at an advantage functional $η(f)$, equal to $\tanh(ε/2)$ under pure DP; for $k\ge2$ duplicated copies the naive $ε\mapsto kε$ bound $\tanh(kε/2)$ is unattainable, the exact constant being a closed-form staircase attained by geometric noisy counting. That cap is attained inside the local score class used in practice, and it is there that the two measures separate: one mechanism is memorized yet unextractable, another fully extractable yet exactly invisible to every loss-based score. The two-sided blind spot this opens for loss-based auditing and unlearning verification survives on billion-parameter models: a reserved-trigger release is recovered verbatim from one prompt while the audits practitioners deploy certify it clean.

cs.CR↗

BiasMap: Leveraging Cross-Attentions to Discover and Mitigate Hidden Social Biases in Text-to-Image Generation

Bias discovery is critical for black-box generative models, especiall text-to-image (TTI) models. Existing works predominantly focus on output-level demographic distributions, which do not necessarily guarantee concept representations to be disentangled post-mitigation. We propose BiasMap, a model-agnostic framework for uncovering latent concept-level representational biases in stable diffusion models. BiasMap leverages cross-attention attribution maps to reveal structural entanglements between demographics (e.g., gender, race) and semantics (e.g., professions), going deeper into representational bias during the image generation. Using attribution maps of these concepts, we quantify the spatial demographics-semantics concept entanglement via Intersection over Union (IoU), offering a lens into bias that remains hidden in existing fairness discovery approaches. In addition, we further utilize BiasMap for bias mitigation through energy-guided diffusion sampling that directly modifies latent noise space and minimizes the expected SoftIoU during the denoising process. Our findings show that existing fairness interventions may reduce the output distributional gap but often fail to disentangle concept-level coupling, whereas our mitigation method can mitigate concept entanglement in image generation while complementing distributional bias mitigation.

cs.CV↗

Algorithmic Recourse in Abnormal Multivariate Time Series

Algorithmic recourse provides actionable recommendations to alter unfavorable predictions of machine learning models, enhancing transparency through counterfactual explanations. While significant progress has been made in algorithmic recourse for static data, such as tabular and image data, limited research explores recourse for multivariate time series, particularly for reversing abnormal time series. This paper introduces Recourse in time series Anomaly Detection (RecAD), a framework for addressing anomalies in multivariate time series using backtracking counterfactual reasoning. By modeling the causes of anomalies as external interventions on exogenous variables, RecAD predicts recourse actions to restore normal status as counterfactual explanations, where the recourse function, responsible for generating actions based on observed data, is trained using an end-to-end approach. Experiments on synthetic and real-world datasets demonstrate its effectiveness.

cs.LG↗

BadSAD: Clean-Label Backdoor Attacks against Deep Semi-Supervised Anomaly Detection

Image anomaly detection (IAD) is essential in applications such as industrial inspection, medical imaging, and security. Despite the progress achieved with deep learning models like Deep Semi-Supervised Anomaly Detection (DeepSAD), these models remain susceptible to backdoor attacks, presenting significant security challenges. In this paper, we introduce BadSAD, a novel backdoor attack framework specifically designed to target DeepSAD models. Our approach involves two key phases: trigger injection, where subtle triggers are embedded into normal images, and latent space manipulation, which positions and clusters the poisoned images near normal images to make the triggers appear benign. Extensive experiments on benchmark datasets validate the effectiveness of our attack strategy, highlighting the severe risks that backdoor attacks pose to deep learning-based anomaly detection systems.

cs.CV↗

Backdoor Attack against One-Class Sequential Anomaly Detection Models

Deep anomaly detection on sequential data has garnered significant attention due to the wide application scenarios. However, deep learning-based models face a critical security threat - their vulnerability to backdoor attacks. In this paper, we explore compromising deep sequential anomaly detection models by proposing a novel backdoor attack strategy. The attack approach comprises two primary steps, trigger generation and backdoor injection. Trigger generation is to derive imperceptible triggers by crafting perturbed samples from the benign normal data, of which the perturbed samples are still normal. The backdoor injection is to properly inject the backdoor triggers to comprise the model only for the samples with triggers. The experimental results demonstrate the effectiveness of our proposed attack strategy by injecting backdoors on two well-established one-class anomaly detection models.

cs.LG↗

LogGPT: Log Anomaly Detection via GPT

Detecting system anomalies based on log data is important for ensuring the security and reliability of computer systems. Recently, deep learning models have been widely used for log anomaly detection. The core idea is to model the log sequences as natural language and adopt deep sequential models, such as LSTM or Transformer, to encode the normal patterns in log sequences via language modeling. However, there is a gap between language modeling and anomaly detection as the objective of training a sequential model via a language modeling loss is not directly related to anomaly detection. To fill up the gap, we propose LogGPT, a novel framework that employs GPT for log anomaly detection. LogGPT is first trained to predict the next log entry based on the preceding sequence. To further enhance the performance of LogGPT, we propose a novel reinforcement learning strategy to finetune the model specifically for the log anomaly detection task. The experimental results on three datasets show that LogGPT significantly outperforms existing state-of-the-art approaches.

cs.LG↗

On Root Cause Localization and Anomaly Mitigation through Causal Inference

Due to a wide spectrum of applications in the real world, such as security, financial surveillance, and health risk, various deep anomaly detection models have been proposed and achieved state-of-the-art performance. However, besides being effective, in practice, the practitioners would further like to know what causes the abnormal outcome and how to further fix it. In this work, we propose RootCLAM, which aims to achieve Root Cause Localization and Anomaly Mitigation from a causal perspective. Especially, we formulate anomalies caused by external interventions on the normal causal mechanism and aim to locate the abnormal features with external interventions as root causes. After that, we further propose an anomaly mitigation approach that aims to recommend mitigation actions on abnormal features to revert the abnormal outcomes such that the counterfactuals guided by the causal mechanism are normal. Experiments on three datasets show that our approach can locate the root causes and further flip the abnormal labels.

cs.LG↗

Robust Fraud Detection via Supervised Contrastive Learning

Deep learning models have recently become popular for detecting malicious user activity sessions in computing platforms. In many real-world scenarios, only a few labeled malicious and a large amount of normal sessions are available. These few labeled malicious sessions usually do not cover the entire diversity of all possible malicious sessions. In many scenarios, possible malicious sessions can be highly diverse. As a consequence, learned session representations of deep learning models can become ineffective in achieving a good generalization performance for unseen malicious sessions. To tackle this open-set fraud detection challenge, we propose a robust supervised contrastive learning based framework called ConRo, which specifically operates in the scenario where only a few malicious sessions having limited diversity is available. ConRo applies an effective data augmentation strategy to generate diverse potential malicious sessions. By employing these generated and available training set sessions, ConRo derives separable representations w.r.t open-set fraud detection task by leveraging supervised contrastive learning. We empirically evaluate our ConRo framework and other state-of-the-art baselines on benchmark datasets. Our ConRo framework demonstrates noticeable performance improvement over state-of-the-art baselines.

cs.CR↗

Achieving Counterfactual Fairness for Anomaly Detection

Ensuring fairness in anomaly detection models has received much attention recently as many anomaly detection applications involve human beings. However, existing fair anomaly detection approaches mainly focus on association-based fairness notions. In this work, we target counterfactual fairness, which is a prevalent causation-based fairness notion. The goal of counterfactually fair anomaly detection is to ensure that the detection outcome of an individual in the factual world is the same as that in the counterfactual world where the individual had belonged to a different group. To this end, we propose a counterfactually fair anomaly detection (CFAD) framework which consists of two phases, counterfactual data generation and fair anomaly detection. Experimental results on a synthetic dataset and two real datasets show that CFAD can effectively detect anomalies as well as ensure counterfactual fairness.

cs.LG↗

Generating Textual Adversaries with Minimal Perturbation

Many word-level adversarial attack approaches for textual data have been proposed in recent studies. However, due to the massive search space consisting of combinations of candidate words, the existing approaches face the problem of preserving the semantics of texts when crafting adversarial counterparts. In this paper, we develop a novel attack strategy to find adversarial texts with high similarity to the original texts while introducing minimal perturbation. The rationale is that we expect the adversarial texts with small perturbation can better preserve the semantic meaning of original texts. Experiments show that, compared with state-of-the-art attack approaches, our approach achieves higher success rates and lower perturbation rates in four benchmark datasets.

cs.CL↗

Robust Unstructured Knowledge Access in Conversational Dialogue with ASR Errors

Performance of spoken language understanding (SLU) can be degraded with automatic speech recognition (ASR) errors. We propose a novel approach to improve SLU robustness by randomly corrupting clean training text with an ASR error simulator, followed by self-correcting the errors and minimizing the target classification loss in a joint manner. In the proposed error simulator, we leverage confusion networks generated from an ASR decoder without human transcriptions to generate a variety of error patterns for model training. We evaluate our approach on the DSTC10 challenge targeted for knowledge-grounded task-oriented conversational dialogues with ASR errors. Experimental results show the effectiveness of our proposed approach, boosting the knowledge-seeking turn detection (KTD) F1 significantly from 0.9433 to 0.9904. Knowledge cluster classification is boosted from 0.7924 to 0.9333 in Recall@1. After knowledge document re-ranking, our approach shows significant improvement in all knowledge selection metrics, from 0.7358 to 0.7806 in Recall@1, from 0.8301 to 0.9333 in Recall@5, and from 0.7798 to 0.8460 in MRR@5 on the test set. In the recent DSTC10 evaluation, our approach demonstrates significant improvement in knowledge selection, boosting Recall@1 from 0.495 to 0.7144 compared to the official baseline. Our source code is released in GitHub https://github.com/yctam/dstc10_track2_task2.git.

cs.CL↗

Fine-grained Anomaly Detection in Sequential Data via Counterfactual Explanations

Anomaly detection in sequential data has been studied for a long time because of its potential in various applications, such as detecting abnormal system behaviors from log data. Although many approaches can achieve good performance on anomalous sequence detection, how to identify the anomalous entries in sequences is still challenging due to a lack of information at the entry-level. In this work, we propose a novel framework called CFDet for fine-grained anomalous entry detection. CFDet leverages the idea of interpretable machine learning. Given a sequence that is detected as anomalous, we can consider anomalous entry detection as an interpretable machine learning task because identifying anomalous entries in the sequence is to provide an interpretation to the detection result. We make use of the deep support vector data description (Deep SVDD) approach to detect anomalous sequences and propose a novel counterfactual interpretation-based approach to identify anomalous entries in the sequences. Experimental results on three datasets show that CFDet can correctly detect anomalous entries.

cs.LG↗

On the Convergence of the Monte Carlo Exploring Starts Algorithm for Reinforcement Learning

A simple and natural algorithm for reinforcement learning (RL) is Monte Carlo Exploring Starts (MCES), where the Q-function is estimated by averaging the Monte Carlo returns, and the policy is improved by choosing actions that maximize the current estimate of the Q-function. Exploration is performed by "exploring starts", that is, each episode begins with a randomly chosen state and action, and then follows the current policy to the terminal state. In the classic book on RL by Sutton & Barto (2018), it is stated that establishing convergence for the MCES algorithm is one of the most important remaining open theoretical problems in RL. However, the convergence question for MCES turns out to be quite nuanced. Bertsekas & Tsitsiklis (1996) provide a counter-example showing that the MCES algorithm does not necessarily converge. Tsitsiklis (2002) further shows that if the original MCES algorithm is modified so that the Q-function estimates are updated at the same rate for all state-action pairs, and the discount factor is strictly less than one, then the MCES algorithm converges. In this paper we make headway with the original and more efficient MCES algorithm given in Sutton & Barto (1998), establishing almost sure convergence for Optimal Policy Feed-Forward MDPs, which are MDPs whose states are not revisited within any episode when using an optimal policy. Such MDPs include a large class of environments such as all deterministic environments and all episodic environments with a timestep or any monotonically changing values as part of the state. Different from the previous proofs using stochastic approximations, we introduce a novel inductive approach, which is very simple and only makes use of the strong law of large numbers.

cs.LG↗

Trustworthy Anomaly Detection: A Survey

Anomaly detection has a wide range of real-world applications, such as bank fraud detection and cyber intrusion detection. In the past decade, a variety of anomaly detection models have been developed, which lead to big progress towards accurately detecting various anomalies. Despite the successes, anomaly detection models still face many limitations. The most significant one is whether we can trust the detection results from the models. In recent years, the research community has spent a great effort to design trustworthy machine learning models, such as developing trustworthy classification models. However, the attention to anomaly detection tasks is far from sufficient. Considering that many anomaly detection tasks are life-changing tasks involving human beings, labeling someone as anomalies or fraudsters should be extremely cautious. Hence, ensuring the anomaly detection models conducted in a trustworthy fashion is an essential requirement to deploy the models to conduct automatic decisions in the real world. In this brief survey, we summarize the existing efforts and discuss open problems towards trustworthy anomaly detection from the perspectives of interpretability, fairness, robustness, and privacy-preservation.

cs.LG↗