3 ms·
Point 1 draws parallels with machine learning's pixel attacks. That one pixel doesn't define the image, but it happens that changing it changes the classificati
by __s 8y ago
Point 1 draws parallels with machine learning's pixel attacks. That one pixel doesn't define the image, but it happens that changing it changes the classification of the whole
- bambamboom 8y agoPeople who downvoted thid- can you explain? I see this as a very nice simplification of point 1 and am glad that _s brought it up