AI to get X-ray vision: Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory are getting close to creating AI that can see through walls, Geek.com reported. The research team is using AI to analyze radio signals bouncing off human bodies. The result is a neural network-generated stick figure that moves like the targeted person does.
Dr. AI will see you now: Perhaps more useful that looking through walls, some AI technologies are now being used to identify tuberculosis, pneumonia, upper respiratory infection, and bronchitis based on how a cough sounds, said AdWeek. Several companies are exploring other ways to use AI in healthcare settings.
Encryption wars, part 207: Apple has moved to close a security hole that law enforcement agencies used to defeat encryption on iPhones, according to many news reports, including one in the New York Times. The Apple move set off a new round of debate about encrypted devices and law enforcement access, the Washington Post noted.
It appears that at least one company that builds iPhone cracking tools already has a workaround, however, Motherboard reported.
Meanwhile, an FBI official suggested that each encrypted device that law enforcement agencies cannot crack represents a victim without justice, BusinessInsider. Continue reading
Getting ready for your first network admin interview? Here are technical questions to expect, along with the answers.
Its all automated
One of my readers asked me a question that came up in his business strategy class:
Why did routers and switches end up being vertically integrated (the same person makes the hardware and the software)? Why didn't they go down the same horizontal path as compute (with Intel making chips, OEMs making systems and Microsoft providing the OS)? Why did this resemble the pre-Intel model of IBM, DEC, Sun…?
Simple answer: because nobody was interested in disaggregating them.
Read more ...DeepTest: automated testing of deep-neural-network-driven autonomous cars Tian et al., ICSE’18
How do you test a DNN? We’ve seen plenty of examples of adversarial attacks in previous editions of The Morning Paper, but you couldn’t really say that generating adversarial images is enough to give you confidence in the overall behaviour of a model under all operating conditions. Adversarial images approach things from a ‘think like an attacker’ mindset. We want to ‘think like a tester.’ For example, the work on DeepXplore which uses model ensembles to find differences in outputs that suggest bugs. The importance of testing DNNs is especially obvious when it comes to applications such as autonomous driving. Several of the ideas from DeepXplore are used in DeepTest, which looks specifically at testing of autonomous driving system. I think you could apply the DeepTest techniques to test other kinds of DNNs as well.
…despite the tremendous progress, just like traditional software, DNN-based software, including the ones used for autonomous driving, often demonstrate incorrect/unexpected corner-case behaviours that lead to dangerous consequences like a fatal collision.
DeepTest is a system designed to aid in the testing of autonomous driving models. When used to test three of Continue reading
The 7th RIPE South-East Europe (SEE 7) meeting is being held on 18-19 June 2018 in Timisoara, Romania, and is focusing on several of the subjects of interest to the Internet Society. It’s also being chaired by our colleague Jan Žorž, whilst I’ll be talking about IoT Security and the OTA IoT Trust Framework.
In Monday, there are talks on BGP monitoring from Paolo Lucente (pmacct), and from Krzysztof Grzegorz Szarkowicz (Juniper Networks) on improvements to routing protocols to suit the centralised data centre-based architectures that are becoming more prevalent on the Internet, and which are the subject of an Internet Draft. Zoran Perovic (SOX) will also talk about paradigm shifts in the implementation of Internet Exchange Points.
On Tuesday, there will be a discussion led by Goran Slavic (SOX) on implementing MANRS in an IXP, which is very relevant to the current MANRS initiative which is increasingly being adopted by IXPs. Our colleague Jan will then be presenting about RIPE-690 which provides recommendations for IPv6 address prefix assignments for end-users. Preceding this, will be an update on IPv6 adoption in the SEE region from Massimiliano Stucchi (RIPE NCC).
Some other highlights are the talk on Quad9DNS by Nishal Goburdhan (PCH) that’s supporting Continue reading
I hate long support lifecycles for hardware and software. Yes, you should be able to buy a new iPhone or switch and use it for 3+ years. But some people want 10+yrs of support, and wail and moan when vendors end support. This is wrong. It drives up costs & complexity, and makes your systems less robust, not more. It’s a false sense of security. Plan to buy smaller & cheaper, and upgrade frequently.
Vendors don’t want to do long support lifecycles. They will do them, because people pay for it, but there comes a point where they put a line in the sand. “Sorry, that system is now EoL.”
Why?
Costs: Testing software and hardware combinations is hard work. Add many years of released hardware & software combinations, and it gets much harder. More racks of gear & more permutations == more costs.
Complexity: It’s hard enough to test against a small set. But now you have to deal with obscure systems acquired from a third party 7 years ago? Complexity == time and money.
Motivation: Hands up who wants to work on legacy systems? Exactly. It’s hard to motivate engineers to support Continue reading
Yesterday we continued a long tradition at DockerCon, the Cool Hacks closing keynote. In our Cool Hacks keynote, we like to emphasize applications that push the limits and applications that represent major future trends in container workloads. We also like to feature applications that demonstrate how Docker fueled innovation can be used every day.
This DockerCon, the three applications we chose embodied all of these characteristics.
Our first hack, by Christopher Heistand of the Johns Hopkins University Applied Physics Laboratory is helping save the world. The Double Asteroid Redirect Mission Test (DART) is testing kinetic impact against an asteroid to measure whether one can be redirected. They use Docker to emulate the specialized and expensive hardware, saving them money and development time.
David Aronchick (@aronchick ) and Michelle Casbon (@texasmichelle) demonstrated our second hack with Kubeflow. Machine learning in production workloads, at scale.
And finally, Idit Levine (@Idit_Levine) showed us Gloo. Gloo gives you the portability and choice of a serverless framework, from cloud services like AWS Lambda to running one of the several containerized self-hosted serverless frameworks. All running in Docker EE.
Check out our Cool Hacks closing keynote.
And finally, we wrapped up inviting Continue reading