Robustness News
The latest Robustness coverage curated for game developers.
A New Type of Adversarial Examples
Researchers have proposed a new class of adversarial examples that keep a model’s answer unchanged while making the input look very different from the original. The work adds negative variants …
Research
arXiv cs.GR
·
1 month ago
Generalizable Vision-Language Few-Shot Adaptation with Predictive Prompts and Negative Learning
A new few-shot vision-language method claims a 4.61% average gain at 16-shot by only suppressing the classes a query is actually likely to confuse. For game teams using CLIP-like models …
Research
arXiv cs.GR
·
3 months, 4 weeks ago