Ou Wu

Branch Landing (BRL) is a novel forward-edge CFI framework for RISC-V that uses Bloom filters to overcome the source authorization limitations of existing hardware CFI, achieving low overhead for fine-grained source control.

The paper proposes R-CoT, a reasoning-layer watermarking framework that embeds ownership watermarks directly into the stable reasoning path of LLMs, achieving high robustness against perturbations.

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.

GRKV introduces a training-free KV-cache merging method that uses global regression to distribute information from evicted tokens, solving the over-merging problem inherent in span-based retention.

AlphaToken is a novel response token valuation framework that improves LLM post-training by decoupling token selection into task-specific adaptation and stability preservation, leading to better performance and reduced catastrophic forgetting.

The paper proposes Joint Neighborhood Optimization (JNO), a novel knowledge-editing framework that jointly addresses the coupled pressures of desirable knowledge propagation and unintended knowledge leakage during single-edit updates in LLMs.