Adversarial ML Problems Are Getting Harder to Solve and to Evaluate

J Rando, J Zhang, N Carlini, F Tramèr - arxiv preprint arxiv:2502.02260, 2025 - arxiv.org
In the past decade, considerable research effort has been devoted to securing machine
learning (ML) models that operate in adversarial settings. Yet, progress has been slow even …

Multi-Faceted Studies on Data Poisoning can Advance LLM Development

P He, Y **ng, H Xu, Z **ang, J Tang - arxiv preprint arxiv:2502.14182, 2025 - arxiv.org
The lifecycle of large language models (LLMs) is far more complex than that of traditional
machine learning models, involving multiple training stages, diverse data sources, and …