Yong Liu

This paper develops improved Gaussian mechanisms for Rényi Pufferfish Privacy (RPP) by incorporating Gaussian and Gaussian-mixture priors, significantly reducing the required noise and improving the privacy-utility trade-off.

The paper introduces SHADOWMERGE, a novel poisoning attack that successfully compromises graph-based agent memory by exploiting relation-channel conflicts, achieving a high attack success rate across multiple real-world benchmarks.

This paper introduces Dependency Steering, a novel attack paradigm demonstrating that malicious agent skills can actively bias LLM coding agents to use attacker-controlled packages, posing a significant, hard-to-detect software supply chain risk.

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex, open-world agentic scenarios.

BlockBatch introduces a novel framework that efficiently accelerates diffusion language model (dLLM) inference by simultaneously executing multiple block-size branches for a single request, achieving significant speedup while maintaining accuracy.

The paper introduces AgentDoG 1.5, a lightweight and scalable alignment framework that significantly improves AI agent safety and security for complex open-world agent deployments.

Lumos-Nexus is a training-efficient framework that enhances video generation quality by progressively bridging generation from a lightweight model to a high-fidelity generator in a shared latent space, without sacrificing reasoning capabilities.

The paper introduces Confidence-Adaptive SwiGLU ($κ$-SwiGLU), a novel gating mechanism for Mixture-of-Experts (MoE) models that dynamically adjusts the gate sharpness based on token-level routing confidence, improving performance with minimal overhead.