jndsn402
Joined 12 years ago
Joined 12 years ago
Myth: Deep learning is more vulnerable to adversarial examples than other kind of machine learning.
Fact: So far we have been able to generate adversarial examples for every model we have tested, including simple traditional machine learning models like nearest neighbor. Deep learning with adversarial training is the most resistant technique we have studied so far.
But the larger point is that adversarial examples are just one demonstration that algorithms are still quite primitive. Nobody should be killed on the basis of a statistical algorithm alone (and to be fair, the article does not show that anyone has been).For instance, maybe your sampling ignored a particular region of the country or socioeconomic strata etc.
As a consumer though, augmented reality ( eg Hololens and Magic Leap) seems like it is about to become very big, and probably utilizes machine vision.
Can you recommend some resources for learning about web scraping?
Your description of how insurance makes money is the standard understanding. However, I would think that most products do have some underwriting gain built in as well.
It can get pretty complex as you need to set up a liability (reserve) to match the assets (premium) that come in. Interest causes both the asset and liability to increase. As benefits are paid out, both the asset and liability decrease. If the liability decreases more than the asset, you make money. Or if you have a substantial surplus (asset - liability), the interest on that is pure profit as well.