Skip to main content

Breaking neural networks with adversarial attacks - Towards Data Science

xxx

"First, as we saw above, it’s easy to attain high confidence in the incorrect classification of an adversarial example — recall that in the first ‘panda’ example we looked at, the network is less sure of an actual image looking like a panda (57.7%) than our adversarial example on the right looking like a gibbon (99.3%). Another intriguing point is how imperceptibly little noise we needed to add to fool the system — after all, clearly, the added noise is not enough to fool us, the humans."

From "Breaking neural networks with adversarial attacks - Towards Data Science".

xxx

Comments

Post a Comment

Popular posts from this blog

new survey results from USA Technologies (USAT) suggest. The payments technology company, which enables electronic payments for self-service machines, compared consumer spending activity at 35 of its vending machines in urban areas with a high concentration of iPhone users between week one and week four of the installation of new digital signage promoting Apple Pay. Consumers made more contactless purchases: The vending machine providers saw an average contactless transaction revenue increase of 89%, implying that consumers are making more mobile payments. This was likely driven by the clear advertisement of Apple Pay. Total transactions increased: Consumers made more purchases overall. This was probably driven by an uptick in mobile payments, given the major increase in contactless transaction revenue. And the vending machines may have attracted many first-time users who were drawn in by the digital advertising of Apple Pay. Boosting awareness is key to unlocking pent-up demand amon...
Post a Comment
Read more

David | LinkedIn

Alex Todd rather kindly called this a “most lucid explanation of digital identity management” I would highly recommend this presentation to anyone interested in understanding prospective blockchain based identity architectures From David | LinkedIn This is very kind, but as I said in the link, this is thinking out loud and far from a fully-developed solution. We’re working on parts of this for different clients and I can see that there is something there - a genuinely new way of solving some old problems - but it’s early days.
Post a Comment
Read more

Crystal clear: Creating hyper-transparent property markets - JLL Real Views

xxx The issue of transparency is going to play a major role in the allocation of an avalanche of global commercial investment into real estate over the next decade. With total sums at stake predicted to rise from US$700 billion to over $1 trillion a year by the mid 2020s, according to JLL, countries and cities which meet investor demands on transparency will be best placed to attract these capital flows. From Crystal clear: Creating hyper-transparent property markets - JLL Real Views xxx
Post a Comment
Read more