arXiv subjects

Xiangtian Li

Publications and source records attributed to Xiangtian Li.

13 recordsLinked to original sources

arXiv2026

Subliminal Learning as Trait-Direction Drift: A Mechanism and Targeted Control under SFT Distillation

Beyond intended capabilities, model distillation can transfer hidden traits from a teacher. A teacher biased by a system prompt can generate semantically clean training data, such as numeric sequences, that still causes a downstream student to inherit the hidden preference, a phenomenon known as subliminal learning. Prior work has identified several parts of this process. How the signal builds up during training and produces behavioral transfer remains unclear, making targeted mitigation difficult. We propose and validate trait-direction drift as a mechanism for subliminal learning: biased generation creates measurable preference gaps in teacher data, and student-recognizable gaps induce trait-aligned updates during supervised fine-tuning that accumulate into behavioral transfer. Guided by this mechanism, we propose probe-space corridor regularization, a targeted defense that constrains drift along a calibrated trait direction during distillation. The method substantially reduces hidden-trait transfer, preserving task performance: for example, it lowers malicious-response transfer from 29.55% to 6.45% with low main-task accuracy cost, and consistently suppresses animal-preference transfer across the main Qwen setting. The preference-gap, training-trajectory, and intervention evidence links subliminal learning to trait-direction drift and motivates corridor regularization as a targeted control during distillation.

cs.LG↗

arXiv2026

Scaling Model-Generated Distillation Data Can Make Latent Teacher Traits More Recoverable

Scaling model-generated data is usually viewed as improving distillation: more examples should increase coverage, reduce noise, and produce stronger students. We show a second effect: larger datasets can make subtle teacher-specific signals easier to detect in the trained student, even when examples are off-task and never mention the trait. In a controlled setup inspired by subliminal learning, a teacher induced to express a target trait generates restricted off-task data, such as number-only completions. Students trained on different amounts of independent off-task data are evaluated in a separate domain, with matched no-trait controls isolating target-specific transfer. Our main finding is that larger independent datasets make the teacher's induced trait stand out more clearly in the student's later behavior. Other plausible traits may also strengthen with scale, but the target usually grows more. When the small-scale student already favors the target, scaling mainly amplifies that behavior; when it favors a related or salient alternative, more data can shift behavior toward the intended trait. Analyses of learned LoRA updates show a parallel trend. These effects appear across model families, trait types, multi-trait settings, and cross-model transfer. Our results suggest that scaling generated distillation data should be paired with trait-aware curation and evaluation, even when the data appears off-task or benign.

cs.LG↗

arXiv2026

Certified Learning and Equilibrium Implementation under Opaque Partial Commitment

As an extension of existing Bayesian persuasion framework with inadequate message mechanism, we study direct recommendation when a sender is bound by an installed information policy only with probability $\rho$, the realization of binding is hidden, and the receiver does not observe the persistent structural environment. The receiver first sees a payoff-neutral, nonmanipulable calibration sample and then faces a fresh, non-certified deployment interaction. In common, the calibration law identifies only the receiver-facing reduced form, not the latent binding and discretionary kernels. We characterize type-wise $\rho$-implementability, construct the receiver's posterior over the full deployment node, and prove a static direct-following implementation theorem. After every calibration history that passes a posterior-predictive obedience test, the deployment assessment is an exact perfect Bayesian equilibrium: Bayes consistency, receiver sequential rationality, sender sequential rationality, and off-path completion are all verified. Under finite-type separation, common recommendation support, and a positive obedience margin, the test activates such an equilibrium with high probability. Our results keep statistical failure probability distinct from equilibrium approximation. Finally, we embed the original robust value frontier, support-wise linear-programming algorithm, and binary-action fractional-knapsack specialization into this implementation framework

cs.GT↗

arXiv2026

REFLECTOR: Internalizing Step-wise Reflection against Indirect Jailbreak

While Large Language Models (LLMs) demonstrate remarkable capabilities, they remain susceptible to sophisticated, multi-step jailbreak attacks that circumvent conventional surface-level safety alignment by exploiting the internal generation process. To address these vulnerabilities, we propose Reflector, a principled two-stage framework that internalizes self-reflection within the generation trajectory. Reflector first leverages teacher-guided generation to produce high-quality reflection data for supervised fine-tuning (SFT), establishing structured reflection patterns. It subsequently uses Reinforcement Learning (RL) with outcome-driven and reward-validity supervision to instill robust, autonomous self-reflection capabilities. Empirical results show that Reflector achieves Defense Success Rates (DSR) exceeding 90% against complex indirect attacks while generalizing robustly across diverse threat scenarios. Notably, the framework enhances both task-specific and general utility, yielding a 5.85% gain on GSM8K alongside improved performance on knowledge-intensive benchmarks. By internalizing trajectory-level safety, Reflector overcomes the fundamental limitations of surface alignment without significant computational overhead, offering an efficient and scalable solution for the development of safe and capable LLMs.

cs.LG↗

arXiv2026

TrinityGuard: A Unified Framework for Safeguarding Multi-Agent Systems

With the rapid development of LLM-based multi-agent systems (MAS), their significant safety and security concerns have emerged, which introduce novel risks going beyond single agents or LLMs. Despite attempts to address these issues, the existing literature lacks a cohesive safeguarding system specialized for MAS risks. In this work, we introduce TrinityGuard, a comprehensive safety evaluation and monitoring framework for LLM-based MAS, grounded in the OWASP standards. Specifically, TrinityGuard encompasses a three-tier fine-grained risk taxonomy that identifies 20 risk types, covering single-agent vulnerabilities, inter-agent communication threats, and system-level emergent hazards. Designed for scalability across various MAS structures and platforms, TrinityGuard is organized in a trinity manner, involving an MAS abstraction layer that can be adapted to any MAS structures, an evaluation layer containing risk-specific test modules, alongside runtime monitor agents coordinated by a unified LLM Judge Factory. During Evaluation, TrinityGuard executes curated attack probes to generate detailed vulnerability reports for each risk type, where monitor agents analyze structured execution traces and issue real-time alerts, enabling both pre-development evaluation and runtime monitoring. We further formalize these safety metrics and present detailed case studies across various representative MAS examples, showcasing the versatility and reliability of TrinityGuard. Overall, TrinityGuard acts as a comprehensive framework for evaluating and monitoring various risks in MAS, paving the way for further research into their safety and security.

cs.CR↗

arXiv2026

Native Reasoning Models: Training Language Models to Reason on Unverifiable Data

The prevailing paradigm for training large reasoning models--combining Supervised Fine-Tuning (SFT) with Reinforcement Learning with Verifiable Rewards (RLVR)--is fundamentally constrained by its reliance on high-quality, human-annotated reasoning data and external verifiers. This dependency incurs significant data-collection costs, risks embedding human cognitive biases, and confines the reinforcement learning stage to objectively assessable domains like mathematics and coding, leaving a wide range of unverifiable tasks beyond its scope. To overcome these limitations, we introduce NRT (Native Reasoning Training), a novel framework that cultivates complex reasoning by having the model generate its own reasoning traces using only standard question-answer pairs, thereby obviating the need for expert-written demonstrations. NRT reframes the training problem by treating the reasoning process as a latent variable. It employs a unified training objective that models reasoning as an optimization problem, intrinsically rewarding paths that increase the model's likelihood of producing the ground-truth answer. This unified perspective allows us to analyze intrinsic failure modes of prior methods, such as policy collapse, and systematically design more robust reward aggregation functions, creating a self-reinforcing feedback loop where the model learns to think in ways that resolve its own uncertainty. Empirical evaluation on Llama and Mistral model families demonstrates that NRT achieves state-of-the-art performance among verifier-free methods, significantly outperforming standard SFT baselines and prior verifier-free RL methods. Our approach yields particularly strong performance gains in complex reasoning domains and exhibits high robustness to policy collapse, offering a general, scalable path toward building more powerful and broadly applicable reasoning systems.

cs.LG↗

arXiv2025

SafeWork-R1: Coevolving Safety and Intelligence under the AI-45$^{\circ}$ Law

We introduce SafeWork-R1, a cutting-edge multimodal reasoning model that demonstrates the coevolution of capabilities and safety. It is developed by our proposed SafeLadder framework, which incorporates large-scale, progressive, safety-oriented reinforcement learning post-training, supported by a suite of multi-principled verifiers. Unlike previous alignment methods such as RLHF that simply learn human preferences, SafeLadder enables SafeWork-R1 to develop intrinsic safety reasoning and self-reflection abilities, giving rise to safety `aha' moments. Notably, SafeWork-R1 achieves an average improvement of $46.54\%$ over its base model Qwen2.5-VL-72B on safety-related benchmarks without compromising general capabilities, and delivers state-of-the-art safety performance compared to leading proprietary models such as GPT-4.1 and Claude Opus 4. To further bolster its reliability, we implement two distinct inference-time intervention methods and a deliberative search mechanism, enforcing step-level verification. Finally, we further develop SafeWork-R1-InternVL3-78B, SafeWork-R1-DeepSeek-70B, and SafeWork-R1-Qwen2.5VL-7B. All resulting models demonstrate that safety and capability can co-evolve synergistically, highlighting the generalizability of our framework in building robust, reliable, and trustworthy general-purpose AI.

cs.AI↗

arXiv2024

Detecting and Classifying Defective Products in Images Using YOLO

With the continuous advancement of industrial automation, product quality inspection has become increasingly important in the manufacturing process. Traditional inspection methods, which often rely on manual checks or simple machine vision techniques, suffer from low efficiency and insufficient accuracy. In recent years, deep learning technology, especially the YOLO (You Only Look Once) algorithm, has emerged as a prominent solution in the field of product defect detection due to its efficient real-time detection capabilities and excellent classification performance. This study aims to use the YOLO algorithm to detect and classify defects in product images. By constructing and training a YOLO model, we conducted experiments on multiple industrial product datasets. The results demonstrate that this method can achieve real-time detection while maintaining high detection accuracy, significantly improving the efficiency and accuracy of product quality inspection. This paper further analyzes the advantages and limitations of the YOLO algorithm in practical applications and explores future research directions.

cs.CV↗

arXiv2024

DTSGAN: Learning Dynamic Textures via Spatiotemporal Generative Adversarial Network

Dynamic texture synthesis aims to generate sequences that are visually similar to a reference video texture and exhibit specific stationary properties in time. In this paper, we introduce a spatiotemporal generative adversarial network (DTSGAN) that can learn from a single dynamic texture by capturing its motion and content distribution. With the pipeline of DTSGAN, a new video sequence is generated from the coarsest scale to the finest one. To avoid mode collapse, we propose a novel strategy for data updates that helps improve the diversity of generated results. Qualitative and quantitative experiments show that our model is able to generate high quality dynamic textures and natural motion.

cs.CV↗

arXiv2024

Real-time Video Target Tracking Algorithm Utilizing Convolutional Neural Networks (CNN)

Thispaperaimstoresearchandimplementa real-timevideotargettrackingalgorithmbasedon ConvolutionalNeuralNetworks(CNN),enhancingthe accuracyandrobustnessoftargettrackingincomplex scenarios.Addressingthelimitationsoftraditionaltracking algorithmsinhandlingissuessuchastargetocclusion,morphologicalchanges,andbackgroundinterference,our approachintegratestargetdetectionandtrackingstrategies.It continuouslyupdatesthetargetmodelthroughanonline learningmechanismtoadapttochangesinthetarget's appearance.Experimentalresultsdemonstratethat,when dealingwithsituationsinvolvingrapidmotion,partial occlusion,andcomplexbackgrounds,theproposedalgorithm exhibitshighertrackingsuccessratesandlowerfailurerates comparedtoseveralmainstreamtrackingalgorithms.This studysuccessfullyappliesCNNtoreal-timevideotarget tracking,improvingtheaccuracyandstabilityofthetracking algorithmwhilemaintaininghighprocessingspeeds,thus meetingthedemandsofreal-timeapplications.Thisalgorithm isexpectedtoprovidenewsolutionsfortargettrackingtasksin videosurveillanceandintelligenttransportationdomains.

cs.CV↗

arXiv2024

Mitigating Knowledge Conflicts in Language Model-Driven Question Answering

In the context of knowledge-driven seq-to-seq generation tasks, such as document-based question answering and document summarization systems, two fundamental knowledge sources play crucial roles: the inherent knowledge embedded within model parameters and the external knowledge obtained through context. Recent studies revealed a significant challenge: when there exists a misalignment between the model's inherent knowledge and the ground truth answers in training data, the system may exhibit problematic behaviors during inference, such as ignoring input context, or generating unfaithful content. Our investigation proposes a strategy to minimize hallucination by building explicit connection between source inputs and generated outputs. We specifically target a common hallucination pattern in question answering, examining how the correspondence between entities and their contexts during model training influences the system's performance at inference time.

cs.CL↗

arXiv2024

Artistic Neural Style Transfer Algorithms with Activation Smoothing

The works of Gatys et al. demonstrated the capability of Convolutional Neural Networks (CNNs) in creating artistic style images. This process of transferring content images in different styles is called Neural Style Transfer (NST). In this paper, we re-implement image-based NST, fast NST, and arbitrary NST. We also explore to utilize ResNet with activation smoothing in NST. Extensive experimental results demonstrate that smoothing transformation can greatly improve the quality of stylization results.

cs.CV↗

arXiv2024

YOLO-PPA based Efficient Traffic Sign Detection for Cruise Control in Autonomous Driving

It is very important to detect traffic signs efficiently and accurately in autonomous driving systems. However, the farther the distance, the smaller the traffic signs. Existing object detection algorithms can hardly detect these small scaled signs.In addition, the performance of embedded devices on vehicles limits the scale of detection models.To address these challenges, a YOLO PPA based traffic sign detection algorithm is proposed in this paper.The experimental results on the GTSDB dataset show that compared to the original YOLO, the proposed method improves inference efficiency by 11.2%. The mAP 50 is also improved by 93.2%, which demonstrates the effectiveness of the proposed YOLO PPA.

cs.CV↗