AI & Robotics

AI and machine learning algorithms are vulnerable to adversarial samples that have alterations from the originals. That’s especially problematic as natural language models become capable of generating humanlike text, because of their attractiveness to malicious actors who would use them to produce misleading media. In pursuit of a technique that illustrates the extent to which adversarial text…

In AI and machine learning systems, knowledge is typically distilled by training a small model — a student — to mimic a large and cumbersome model — a teacher. The idea is to compress the teacher’s knowledge by using its outputs as labels to optimize the student, but there’s no guarantee knowledge will be transferred to the student when the teacher is considerably large. That’s why a…

Worlds, whose AI helps large organizations observe their physical spaces to ensure security, safety, and productivity, has raised $10 million in a first round of funding. The Dallas-based startup uses multiple cameras to track spaces in 3D over time, providing a more reliable view of what’s happening than typical 2D security video would. Worlds then trains its AI models to find unusual patterns…