qeabc

Joined 11 years ago

Karma

Submissions

Member since

Dec 22

Recent submissions

Perhaps someone who knows the topic can say whether this mathematical phenomena of "adversarial examples" for neural networks can be translated to random forests: http://www.kdnuggets.com/2015/07/deep-learning-adversarial-e...

Barely perceptible changes to an image cause it to be misclassified by neural networks which never saw the image before (this is important, because it rules out simple overfitting). As a non-expert, this suggests to me that, at least for some algorithms, the reasoning "statistical algorithm estimates 99.999% probability of guilt" implies "guilty beyond a reasonable doubt" is unreliable at best.

There is great potential here for safe, effective government if machine learning output is only ever input for humans with common sense, life experience and the ability to interact with the world.

Comment · 10 years ago

> So there is really nothing at all to suggest that those algorithms alone are used to create kill lists.

The first line of the article actually suggests this: 'In 2014, the former director of both the CIA and NSA proclaimed that "we kill people based on metadata."'

Also, even if something does not make sense does not mean that it does not seem to make sense to a particular person or organization.

Comment · 11 years ago

Some comments about how GCC could have avoided this made me think that this is a setback for free software, but why isn't this just a rising tide benefiting both free and proprietary software?

The NCSA license explicitly allows sublicensing, so both the GNU project and Microsoft can create derivatives under free and proprietary licenses respectively, for example.

[I am not a lawyer, but this is the reasoning of the accepted answer to a stackexchange question applied to the NCSA license: http://programmers.stackexchange.com/questions/105912/can-yo...]