arXiv subjects

Shu Hong

Publications and source records attributed to Shu Hong.

8 recordsLinked to original sources

arXiv2026

Localizing Credit at the Divergence: Path-Conditioned Self-Distillation for LLM Reasoning

Reinforcement learning from verifiable rewards assigns a single scalar to each rollout, leaving token-level credit assignment underspecified in long reasoning traces. On-policy self-distillation addresses this by letting the same model act as a teacher conditioned on privileged information, producing a dense per-token signal. But the common choice of a ground-truth answer is only an endpoint cue: on terse-answer tasks, the teacher falls silent at the intermediate positions where path-level guidance matters most. We propose Hindsight Self-Distillation (HSD), which conditions the teacher on a successful peer rollout drawn from the current training group. Such a peer is an exact sample from the success-conditioned policy, requiring no additional sampled rollouts. By providing a full successful continuation rather than only the final answer, the resulting credit signal concentrates at the divergence position between a failed rollout and a successful peer. Across Qwen3-8B and Qwen3-32B on math and code benchmarks, HSD obtains the best result against GRPO variants and on-policy distillation baselines, with the largest gains on terse-answer tasks such as AIME.

cs.LG↗

arXiv2026

ZODIAC: Zero-shot Offline Diffusion for Inferring Multi-xApps Conflicts in Open Radio Access Networks

Open Radio Access Network (O-RAN) enables network control through multi-vendor xApps operating both within and across layers, subnets, and domains, whose concurrent execution can trigger conflicts that are latent during the development phase. Existing conflict management approaches rely heavily on joint-execution data, which is often unavailable in practice. To address this limitation, we formalize a novel problem termed conflict reasoning, which involves identifying conflict-inducing conditions given only marginal datasets from each individual xApp. We propose ZODIAC, a three-stage framework for zero-shot conflict condition inference that comprises uncertainty-aware surrogate model training, trajectory-level diffusion training, and compositional guided denoising for efficient, physics-constrained, and reliable condition search. We derive a theoretical lower confidence bound showing that the compositional reasoning in ZODIAC serves as a principled surrogate for true conflict severity, with the epistemic penalty directly controlling the approximation gap. We evaluate ZODIAC on both the lightweight Mobile-Env platform across all three O-RAN Alliance conflict types (direct, indirect, and implicit) and a realistic NS-O-RAN-Flexric simulator. ZODIAC consistently outperforms baseline condition search methods, achieving over 20% higher True Positive Rate at Top-20, substantially stronger Spearman rank correlation, greater scenario diversity, and competitive computational efficiency. Ablation studies confirm the necessity of each guidance component, with epistemic uncertainty penalties proving essential for filtering spurious conflicts. To the best of our knowledge, ZODIAC is the first framework in O-RAN that enables conflict reasoning from marginal offline data without requiring any joint-execution traces.

cs.NI↗

arXiv2025

Global Optimization on Graph-Structured Data via Gaussian Processes with Spectral Representations

Bayesian optimization (BO) is a powerful framework for optimizing expensive black-box objectives, yet extending it to graph-structured domains remains challenging due to the discrete and combinatorial nature of graphs. Existing approaches often rely on either full graph topology-impractical for large or partially observed graphs-or incremental exploration, which can lead to slow convergence. We introduce a scalable framework for global optimization over graphs that employs low-rank spectral representations to build Gaussian process (GP) surrogates from sparse structural observations. The method jointly infers graph structure and node representations through learnable embeddings, enabling efficient global search and principled uncertainty estimation even with limited data. We also provide theoretical analysis establishing conditions for accurate recovery of underlying graph structure under different sampling regimes. Experiments on synthetic and real-world datasets demonstrate that our approach achieves faster convergence and improved optimization performance compared to prior methods.

cs.LG↗

arXiv2025

Perception Graph for Cognitive Attack Reasoning in Augmented Reality

Augmented reality (AR) systems are increasingly deployed in tactical environments, but their reliance on seamless human-computer interaction makes them vulnerable to cognitive attacks that manipulate a user's perception and severely compromise user decision-making. To address this challenge, we introduce the Perception Graph, a novel model designed to reason about human perception within these systems. Our model operates by first mimicking the human process of interpreting key information from an MR environment and then representing the outcomes using a semantically meaningful structure. We demonstrate how the model can compute a quantitative score that reflects the level of perception distortion, providing a robust and measurable method for detecting and analyzing the effects of such cognitive attacks.

cs.AI↗

arXiv2025

AnalogFed: Privacy-Preserving Discovery of Analog Circuits at Scale with Federated Generative AI

Recent advances in generative AI (GenAI) have shown transformative potential for modern hardware design. However, existing GenAI-driven approaches fall short of enabling large-scale electronic design automation (EDA) due to the proprietary and siloed nature of hardware datasets, which cannot be centralized for model training. Achieving at-scale GenAI-driven EDA, therefore, requires a novel privacy-preserving framework that can leverage distributed data without compromising confidentiality. This work introduces AnalogFed, the first privacy-preserving framework for large-scale analog circuit topology discovery using federated learning (FedL) and GenAI. AnalogFed establishes the feasibility of collaborative analog topology design while addressing key security challenges: it mitigates membership inference attacks (MIAs) through a novel input perturbation strategy based on dummy token injection, and defends against model inversion attacks with customized, efficient homomorphic encryption. Extensive experiments demonstrate AnalogFed's effectiveness and efficiency, achieving strong privacy protection without degrading model utility. This framework lays the foundation for scalable, multi-party collaboration in next-generation hardware design automation with GenAI.

cs.LG↗

arXiv2024

Lightweight Federated Learning with Differential Privacy and Straggler Resilience

Federated learning (FL) enables collaborative model training through model parameter exchanges instead of raw data. To avoid potential inference attacks from exchanged parameters, differential privacy (DP) offers rigorous guarantee against various attacks. However, conventional methods of ensuring DP by adding local noise alone often result in low training accuracy. Combining secure multi-party computation (SMPC) with DP, while improving the accuracy, incurs high communication and computation overheads as well as straggler vulnerability, in either client-to-server or client-to-client links. In this paper, we propose LightDP-FL, a novel lightweight scheme that ensures provable DP against untrusted peers and server, while maintaining straggler resilience, low overheads and high training accuracy. Our scheme incorporates both individual and pairwise noise into each client's parameter, which can be implemented with minimal overheads. Given the uncertain straggler and colluder sets, we utilize the upper bound on the numbers of stragglers and colluders to prove sufficient noise variance conditions to ensure DP in the worst case. Moreover, we optimize the expected convergence bound to ensure accuracy performance by flexibly controlling the noise variances. Using the CIFAR-10 dataset, our experimental results demonstrate that LightDP-FL achieves faster convergence and stronger straggler resilience compared to baseline methods of the same DP level.

cs.CR↗

arXiv2023

Regulating Clients' Noise Adding in Federated Learning without Verification

In federated learning (FL), clients cooperatively train a global model without revealing their raw data but gradients or parameters, while the local information can still be disclosed from local outputs transmitted to the parameter server. With such privacy concerns, a client may overly add artificial noise to his local updates to compromise the global model training, and we prove the selfish noise adding leads to an infinite price of anarchy (PoA). This paper proposes a novel pricing mechanism to regulate privacy-sensitive clients without verifying their parameter updates, unlike existing privacy mechanisms that assume the server's full knowledge of added noise. Without knowing the ground truth, our mechanism reaches the social optimum to best balance the global training error and privacy loss, according to the difference between a client's updated parameter and all clients' average parameter. We also improve the FL convergence bound by refining the aggregation rule at the server to account for different clients' noise variances. Moreover, we extend our pricing scheme to fit incomplete information of clients' privacy sensitivities, ensuring their truthful type reporting and the system's ex-ante budget balance. Simulations show that our pricing scheme greatly improves the system performance especially when clients have diverse privacy sensitivities.

cs.GT↗

arXiv2023

Location Privacy Protection Game against Adversary through Multi-user Cooperative Obfuscation

In location-based services(LBSs), it is promising for users to crowdsource and share their Point-of-Interest(PoI) information with each other in a common cache to reduce query frequency and preserve location privacy. Yet most studies on multi-user privacy preservation overlook the opportunity of leveraging their service flexibility. This paper is the first to study multiple users' strategic cooperation against an adversary's optimal inference attack, by leveraging mutual service flexibility. We formulate the multi-user privacy cooperation against the adversary as a max-min adversarial game and solve it in a linear program. Unlike the vast literature, even if a user finds the cached information useful, we prove it beneficial to still query the platform to further confuse the adversary. As the linear program's computational complexity still increases superlinearly with the number of users' possible locations, we propose a binary obfuscation scheme in two opposite spatial directions to achieve guaranteed performance with only constant complexity. Perhaps surprisingly, a user with a greater service flexibility should query with a less obfuscated location to add confusion. Finally, we provide guidance on the optimal query sequence among LBS users. Simulation results show that our crowdsourced privacy protection scheme greatly improves users' privacy as compared with existing approaches.

cs.CR↗