markchen90

6 anni fadiscuss

I didn't notice any obvious visual differences, but I'm also not an expert on adversarial examples. The transformer models were similarly susceptible to attacks, but while adversarial examples transferred well within a model class (~40%), they did not across model classes (~5%). These are rough numbers from memory, don't hold me accountable!

markchen90

6 anni fadiscuss

Author here. You're absolutely right that "understanding" is a fuzzy word. As you pointed out, part of reason we hold this belief is that the model can generate diverse samples and successfully complete out-of-distribution inputs. But the other part is that the model learns (without labels) useful features for classifying objects. Would be very interesting to test it on a broader set of datasets which measure other 2D image characteristics.

markchen90

6 anni fadiscuss

Author here. I ran some early experiments a while ago, and it looked like adversarial examples for convnet classifiers didn't transfer to transformer classifiers and vice versa. Definitely worth looking more into!

markchen90

6 anni fadiscuss

I don't think this is true. The ResNet was born at Microsoft, DQN was born at Deepmind, the Transformer was born at Google, and GPT2 was born at OpenAI.

I'm obviously biased since I work at an industry AI lab, but we both have important roles to play.

comments