Not all types of materials are equally studied, which can lead to bias when using past data find new options.

Large pre-trained language models are successfully being used in a variety of tasks, across many languages. With this ever-increasing usage, the risk harmful side effects also rises, for example by reproducing and reinforcing stereotypes. However, detecting mitigating these harms is difficult to do general becomes computationally expensive when tackling multiple languages or considering differe...

Many techniques have been proposed to harden programs with protection mechanisms defend against vulnerability exploits. Unfortunately the vast majority of them cannot be applied closed source software because they require access program code. This paper presents our work on automatically hardening binary code security workarounds, a mechanism that prevents vulnerabilities from being triggered b...