影响力指数
论文质量、代表作、近期表现、广度与样本量置信度综合计算
-/100
发表论文3 篇
平均评分
年均产出3.0 篇/年
Samuele Poppi
研究方向
Watermarking · AI Security · Constitutional AI · Developmental Interpretability · Scalable Oversight · LLM context inference · Deepfakes · Mechanistic Interpretability · LLMs · AI Safety · T2I models safety · LLM Safety · Jailbreaking · Diffusion Models · VLMs · Machine Unlearning · Responsible AI · Explainable AI
19
Mitigating Watermark Forgery in Generative Models via Randomized Key Selection
ICLR 2026Rejected
6
Improving LLM First-Token Predictions in Multiple-Choice Question Answering via Output Prefilling
ICLR 2026Withdrawn
三作-1
Open Models Can Silently Undermine Privacy: Context Inference Attacks without Jailbreaks
ICLR 2026Withdrawn
二作