arXiv subjects

Yuqi Qing

Publications and source records attributed to Yuqi Qing.

7 recordsLinked to original sources

arXiv2026

Mitigating Proxy-Induced Traffic Drift in Website Fingerprinting via Model-Agnostic Traffic Tailoring

Website fingerprinting (WF) based on deep learning can effectively identify websites from encrypted traffic. However, users often rely on proxy protocols to bypass censorship, and the diversity of these protocols poses a major challenge, as WF models trained on traffic from one set of protocols perform poorly when evaluated on that from unseen protocols. We attribute this issue to proxy-induced feature drift, where traffic patterns of the same website vary with the proxy protocol, leading to discrepancies that WF models fail to capture and severe performance degradation. To tackle this issue, we propose PA3, a model-agnostic preprocessing framework to analyze and mitigate the proxy-induced drift. PA3 first fingerprints the protocol-specific drift. These fingerprints are then used to tailor the proxied traffic for feature alignment, which mitigates the drift and considerably improves the generalization of WF models on traffic from unseen protocols. Extensive evaluations demonstrate that PA3 substantially enhances generalization on unseen protocols with an average improvement of 0.12 in F1-score (roughly 27% relative), achieving up to a 0.41 absolute gain across models, which narrows the performance gap introduced by the drift. In the best case, PA3 enables WF models to obtain F1-scores above 0.96 on traffic from unseen protocols.

cs.NI↗

arXiv2026

When Experience Becomes Instruction: Trajectory Poisoning in Self-Evolving Agent Skill Systems

Self-evolving skill (SES) systems distill agent trajectories into persistent skills, allowing untrusted experience to become trusted instruction. We introduce PoisonedEvolution, a trajectory-poisoning attack on this promotion process. Our skill-visible black-box attacker can inspect a target skill and contribute bounded evidence, but cannot observe private pools or evolution logic or edit the skill bank. Artifact poisoning requires Inclusion, Evolution Attribution, and Realization. Attribution is the distinctive bottleneck: the target behavior must appear causally useful, recurrent, and generalizable before promotion. We evaluate four representative security-effect families using inert canary specifications. At 10% attacker support, across six mainstream LLM evolvers in SkillClaw, PoisonedEvolution embeds target behaviors in 546/600 trials (91.0% SER). On the structurally different Trace2Skill pipeline at the same ratio, it embeds target behaviors in 369/600 trials (61.5% SER), demonstrating transfer across evolution architectures. In a representative controlled study, three consistent attacker records suffice in a 30-record batch, whereas a single record is much weaker. Ablations identify recurring support, causal framing, and domain-aligned encoding as the main determinants of success. These findings expose evidence promotion as a security boundary for self-evolving agents.

cs.CR↗

arXiv2026

Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies

Self-evolving LLM agent systems, which autonomously update their model parameters, memory, tools, and architectures, introduce a qualitatively new threat landscape in which adversarial influences become permanently encoded, self-amplify across generations, and propagate through populations without sustained attacker access. We present a systematic security and privacy analysis organized around the Module-Lifecycle Attack Surface (MLAS) matrix, which decomposes the attack surface into five functional modules (Brain, Cognitive Resource, Execution, Self-Design, Collective) $\times$ five lifecycle stages (Bootstrap, Propose, Evaluate, Commit, Serve). Analysis of the resulting 25 cells reveals that 17 face critical threats for which no effective partial mitigation. We identify seven cross-cutting amplification effects that interact synergistically and cannot be addressed by securing individual modules in isolation. Comparative case studies of two open-source frameworks demonstrate that evolution-native design activates $3.5\times$ more attack surface cells and achieves a 100% attack persistence rate (40/40 payloads across all CIA+Privacy categories), while co-located security scanners block only 2.5% of attacks. Our findings establish that self-evolution converts every known attack category from session-bounded to lineage-persistent, gives rise to entirely new attack classes, and renders static defenses structurally inadequate, motivating evolution-aware security frameworks and formal verification for self-modifying systems.

cs.CR↗

arXiv2026

EgoBench: An Interactive Egocentric Multimodal Benchmark for Tool-Using Agents

As AI agents increasingly operate in open, real-world environments, they require a deep synergy of multimodal perception, tool invocation with multi-hop reasoning, and dynamic interaction with users. However, existing benchmarks fail to jointly evaluate these capabilities due to challenges in designing strictly coupled multi-capability tasks, simulating natural and task-constrained user feedback, and ensuring objective evaluation of dynamic interaction. To bridge this gap, we introduce EgoBench, the first interactive multimodal benchmark for tool-using agents. EgoBench comprises 1,045 egocentric-video-grounded tasks covering four daily scenarios, along with a user-agent-tool interactive environment for evaluation. We implement a three-stage synergistic pipeline through which each task is designed to enforce the joint application of visual perception and tool-augmented multi-hop reasoning. We additionally develop a multi-agent simulated user within EgoBench to evaluate agents' interaction capabilities, which generates high-fidelity, task-aligned responses to agents. Furthermore, we establish a deterministic joint validation framework that guarantees objective assessment through process-based and result-based equivalence. Benchmarking eight SOTA video-MLLM agents on EgoBench reveals a severe performance ceiling: the best model achieves only 30.62% accuracy in the best-performing scenario, averaging 19.43% across all four scenarios. Finally, we conduct a multi-dimensional error analysis to disentangle failure modes, exposing capability bottlenecks for advancing future AI agents.

cs.AI↗

arXiv2025

Entanglement growth and information capacity in a quasiperiodic system with a single-particle mobility edge

We investigate the quantum dynamics of a one-dimensional quasiperiodic system featuring a single-particle mobility edge (SPME), described by the generalized Aubry-Andr\'e (GAA) model. This model offers a unique platform to study the consequences of coexisting localized and extended eigenstates, which contrasts sharply with the abrupt localization transition in the standard Aubry-Andr\'e model. We analyze the system's response to a quantum quench through two complementary probes: entanglement entropy (EE) and subsystem information capacity (SIC). We find that the SPME induces a smooth crossover in all dynamical signatures. The EE saturation value exhibits a persistent volume-law scaling in the mobility-edge phase, with an entropy density that continuously decreases as the number of available extended states decreases. Complementing this, the SIC profile interpolates between the linear ramp characteristic of extended systems and the information trapping behavior of localized ones, directly visualizing the mixed nature of the underlying spectrum. Our results establish unambiguous dynamical fingerprints of a mobility edge, providing a crucial non-interacting benchmark for understanding information and entanglement dynamics in more complex systems with mixed phases.

quant-ph↗

arXiv2024

Optimal Dynamical Gauge in the Quantum Rabi Model

In this paper, we investigate the gauge dependence of various physical observables in the quantum Rabi model (QRM) under different potential fields, arising from the Hilbert-space truncation of the atomic degree of freedom. We discover that in both the square-well potential and oscillator potential,the optimal gauges for the ground-state energy of the QRM vary with respect to the cavity frequency, with the dipole gauge being optimal in the low-frequency limit and the Coulomb gauge in the high-frequency limit of the cavity frequency. Additionally, for higher energy levels, the optimal gauge asymptotically approaches the dipole gauge. However, for the dynamical quantity out-time-order correlator (OTOC), we find the necessity to introduce an optimal dynamical gauge. We determine the optimal dynamical gauge by minimizing the mean error between the two-level OTOC and the full Hamiltonian one. We expect that this study will contribute to a more profound understanding of the subtle relation between gauge choice and the dynamics of QED systems.

quant-ph↗

arXiv2023

Low-Quality Training Data Only? A Robust Framework for Detecting Encrypted Malicious Network Traffic

Machine learning (ML) is promising in accurately detecting malicious flows in encrypted network traffic; however, it is challenging to collect a training dataset that contains a sufficient amount of encrypted malicious data with correct labels. When ML models are trained with low-quality training data, they suffer degraded performance. In this paper, we aim at addressing a real-world low-quality training dataset problem, namely, detecting encrypted malicious traffic generated by continuously evolving malware. We develop RAPIER that fully utilizes different distributions of normal and malicious traffic data in the feature space, where normal data is tightly distributed in a certain area and the malicious data is scattered over the entire feature space to augment training data for model training. RAPIER includes two pre-processing modules to convert traffic into feature vectors and correct label noises. We evaluate our system on two public datasets and one combined dataset. With 1000 samples and 45% noises from each dataset, our system achieves the F1 scores of 0.770, 0.776, and 0.855, respectively, achieving average improvements of 352.6%, 284.3%, and 214.9% over the existing methods, respectively. Furthermore, We evaluate RAPIER with a real-world dataset obtained from a security enterprise. RAPIER effectively achieves encrypted malicious traffic detection with the best F1 score of 0.773 and improves the F1 score of existing methods by an average of 272.5%.

cs.CR↗