Zeliang Zhang

Training Large Reasoning Models Efficiently via Progressive Thought Encoding

DRIFT: Directional Reasoning Injection for Fine-Tuning MLLMs

AdvEvo-MARL: Shaping Internalized Safety through Adversarial Co-Evolution in Multi-Agent Reinforcement Learning

Harnessing the Computation Redundancy in ViTs to Boost Adversarial Transferability

Rethinking Audio-Visual Adversarial Vulnerability from Temporal and Modality Perspectives

Understanding Model Ensemble in Transferable Adversarial Attack

FLOPS: Forward Learning with OPtimal Sampling

Will the Inclusion of Generated Data Amplify Bias Across Generations in Future Image Classification Models?

Forward Learning with Differential Privacy

Understanding Model Ensemble in Transferable Adversarial Attack

Enhancing Adversarial Transferability Through Exploiting Multiple Randomized Trajectories for Better Global Guidance

Is Forward Gradient an Effective Tool for Explaining Black-box Models?

Towards good practice in boosting the targeted adversarial attack