3 ms·

Problem is, mimicry involves copying essential properties of the target object: color, texture, shape, movement dynamics and so on. It relies on true ambiguity.

Problem is, mimicry involves copying essential properties of the target object: color, texture, shape, movement dynamics and so on. It relies on true ambiguity. Adversarial examples against neural networks (the interesting ones, anyway) involve a combination of insignificant, seemingly random permutations that only work in their totality. That's a very important difference.

munificent 9y ago
> essential properties of the target object They are only "essential" properties according to the limitations of your particular perceptual system. To a bee that can see UV light, a stick bug may look entirely different from an actual stick. An animal that hunts by scent would find them clearly distinct. There is no such thing as "true" ambiguity unless the two objects are actually the same thing in all respects. If they are distinct but appear the same, it's because they are overlapping in some respects but not all.
- romaniv 9y ago
  > They are only "essential" properties according to the limitations of your particular perceptual system. To a bee that can see UV light, a stick bug may look entirely different from an actual stick. An animal that hunts by scent would find them clearly distinct. This is a very confused statement. Fist, since we are talking about image recognition we are - by definition - talking about vision in the spectrum that can be captured by a digital camera and encoded in a typical image format. Second, there definitely is such a thing as essential property. It is a matter of correlation with reality, as well as internal consistency. For example, plants are green and have leaves because of the way they use sunlight. So permuting color of all leaves in the picture is fundamentally different from permuting luminosity of some random pixels.

3 ms·

Problem is, mimicry involves copying essential properties of the target object: color, texture, shape, movement dynamics and so on. It relies on true ambiguity.

by romaniv 9y ago

munificent 9y ago
> essential properties of the target object They are only "essential" properties according to the limitations of your particular perceptual system. To a bee that can see UV light, a stick bug may look entirely different from an actual stick. An animal that hunts by scent would find them clearly distinct. There is no such thing as "true" ambiguity unless the two objects are actually the same thing in all respects. If they are distinct but appear the same, it's because they are overlapping in some respects but not all.
- romaniv 9y ago
  > They are only "essential" properties according to the limitations of your particular perceptual system. To a bee that can see UV light, a stick bug may look entirely different from an actual stick. An animal that hunts by scent would find them clearly distinct. This is a very confused statement. Fist, since we are talking about image recognition we are - by definition - talking about vision in the spectrum that can be captured by a digital camera and encoded in a typical image format. Second, there definitely is such a thing as essential property. It is a matter of correlation with reality, as well as internal consistency. For example, plants are green and have leaves because of the way they use sunlight. So permuting color of all leaves in the picture is fundamentally different from permuting luminosity of some random pixels.