AI Image Recognition Fooled By Single Pixel Change

Adversarial Model Images

According to research, computers can be misled into thinking that an image of a taxi can be a dog by only altering one pixel. These limits developed from the methods that Japanese function in tricking the extensively utilised AI-based image recognition system.

Several of the other scientists tend to now develop `adversarial’ model images to reveal the fragility of some kinds of recognition software. Experts have cautioned that there is no quick and easy means of fixing image recognition system of stopping them from being duped in this manner.

Su Jiawei together with colleagues at Kyushu University, in their research had made small alterations to plenty of the images which were then analysed by extensively utilised AI-based image recognition systems. All the systems that had been tested had been based on a kind of AI known as deep neural networks.

These systems usually tend to learn on being trained with plenty of various examples for the purpose of providing them with an intellect of how objects such as dogs and taxis tend to vary. It was observed by the researchers that altering one pixel in about 74% of the test images made the neural nets mistakenly label what they saw.

Designed – Pixel Based Attacks

A variety of pixel based attacks had been designed by Japanese researchers which had caught all the state-of-the-art image recognitions system that had been investigated. Mr Su from Kyushu leading the research had commented that as far as they were aware there was no data-set or network which is more robust than others.

 Several other research groups all over the world have been now developing `adversarial examples’ which tend to reveal the flaw of these systems according to Anish Athalye from the Massachusetts institute of Technology – MIT who has been dealing with this issue. A specimen made by Mr Athalye together with his team is a 3D printed turtle that one image classification system insists on labelling a rifle.

He informed BBC that more and more real world schemes have begun to incorporate neural networks and tends to be huge concern which these schemes could be possible to destabilize or attack utilising adversarial examples. He stated that though there had been no instances of malicious attacks in real life, the fact that these apparently smart schemes could be deceived with ease was a matter of concern.

Methods of Resisting Adversarial Exploitation

Web giants comprising of Facebook, Amazon as well as Google seems to be known for investigating methods of resisting adversarial exploitation. He stated that it is not some strange `corner case’ and it has been shown in their work that one can have a single object which steadily fools a network over viewpoints, even in the physical world.

He further added that the machine learning community do not tend to comprehend completely what seems to be going on with adversarial examples or why they seem to exist. Learning system established on neural network tends to involve creating links between large numbers of nodes such as nerve cells in a brain.

Analysis involves the network creating plenty of decision regarding what it tends to see and every decision should lead the network nearer to the accurate answer.

Google Introducing Certified Kubernetes

Since its launch three years back, Kubernetes has gained a lot of popularity over the years, with many vendors even developing their own offerings. What makes the Kubernetes software so successful is that it is programmed to run anywhere and its consistency.

There are roughly 60 new Kubernetes platforms and distributions out there.
 

What is a Kubernetes Certification program?

 
One of the main concerns to users of the Kubernetes program is that it should be consistent and reliable anywhere and at any time. This issue is compounded with the fact that there are many vendors out there who are now offering Kubernetes, so consistency over platforms does become a major concern for any client.

In order to address this issue, the Cloud Native Computing Foundation or more commonly known as CNCF in partnership with the Kubernetes community, have recently introduced the beta version of Kubernetes conformance certification program.
 

The Kubernetes Certification Program:

 

With this certificate in mind, vendors are able to reassure their clients that the software they are using conform to standards and will be reliable in any of the platforms they use it on.

Besides conforming to standards, the Kubernetes certification also gives users confidence of a high level of common functionality of the Kubernetes software over all platforms.

Basically clients can be sure that their software will work as expected.

 

How to get the Kubernetes conformation Certificate?

 
Cloud Native Computing Foundation in partnership with the Kubernetes community are requesting all Kubernetes vendors to run the conformance test and then send the results to the CNCF for review and to give the certificate.

All vendors that have been tested and certified during the beta version of the program will have their names published when the actual version of the program is released.

 

Changes to the Kubernetes certificate:

 
Kubernetes certification standards will receive updates as and when there are software developments to the Kubernetes program itself.

This is done in order to ensure that the updated version of the Kubernetes software conforms to updated standards and a new certificate will be issued in effect.

Every vendor has to get recertified each year in order to provide their clients with the reassurance that not only are they getting a software that complies with uniform standards but is also updated with all the latest and newest in the Kubernetes software.

Once the vendors have their certification in hand they will be able to display the Kubernetes certified logo on all marketing and packaging material.
 

Advantages of Kubernetes Certification:

 
One of the main advantages with a Kubernetes certificate is that there will be smoother flow of data between cloud environments and local operations. The workload can also be split over multiple platforms.
Another advantage is that there will be consistent updates and access to support resources and community software.

BankBot Android Malware Sneaks into the Google Play Store

BankBot: Google’s little Sneak

A malware known for stealing people’s bank details by posing as a legitimate bank’s web page has struck hundreds of users yet again.

The malware was spotted on the Google play store in April of this year, with a recurrence in September and then finally it has come around again in November. Each of these times the Bank Bot nuisance has been removed by the Google team.

What essentially is BankBot?

BankBot is malware that first made its appearance in the Google Play store in April of this year. Posing as a legitimate app it fools users into downloading it. BankBot then asks for permission to SMS, bank and other details, all the while appearing to be a legitimate process of downloading the app.

Finally BankBot poses as a bank page when any payment has to be made. This bank page appears as an overlay to the actual one. BankBot even has access to an individual’s SMS app to take care of dual authentication processes which is required for some banks.

Why was it so easy for users to get fooled?

Users of Google’s Play store downloaded a seemingly genuine app known as Crypto Currency Market Prices. This app appears to be the front of the malware. Having a proper appearance and containing all the details it promised to have, it was very easy for users to get fooled by the app.

By having a working app, users were easily taken in by its appearance to doubt any foul practice. Had it been an app which appeared to be full of spam or one that didn’t work well or could not be opened, it would have given users an idea of what the app really was about.

The second point to note was that the malware was removed twice by the Google team with the security protocols being renewed and updated and in spite of all this BankBot found its way back into the system.

Updates to BankBot:

Since its first appearance in April, the BankBot malware has under gone a series of updates. These updates include code obfusication and the ability to bypass Android’s accessibility services.

This coupled with the fact that it was able to penetrate Google’s security protocol is an added way in which the malware has developed over the span of 7 months.

Users affected by BankBot:

The crypto currency app containing the malware was removed from the Google Play store but not before hundreds of users downloaded it to their mobile devices.

As far as Google is concerned, its 1.4 million were safe from the attack. It is unsure how many users were affected when the malware first arrived on the scene in April of this year.

Google Play store is full of apps that can cause serious damage to an individual’s device or in this case to their bank account. It is therefore important to keep an eye out on what you’re downloading. This is not the first time Google has become embroiled in the malware found on its Play store, recently there was news that a fake Whatsapp app was doing the rounds on the Google Play store.

Why Google is Reading Your Google Docs

Most of the users of the online word processing service by Google, known as Google Docs, were in for a surprise on Tuesday when their documents faced a block as a result of violating the policies. Many users were reported saying they lost access to either their wedding vows or academic researches, for example, receiving notifications saying that their documents had violated the terms of service of Google Docs. This error was reported by dozens of users on the product forums of Google.
While Google acted quickly and put the blame on an erroneous update that locked its users out of their own documents by mistake, the whole incident disclosed what many people did not notice, that Google is scrutinising the apparently personal files stored in Google Docs.

While Google was under pressure to keep a better watch on the kind of videos that were being uploaded on YouTube, deleting content that was illegal from its search results and better monitoring of its advertising network, hardly few people were aware that Google Docs was also being watched. Google Docs is not only used by people for their day-to-day works but also by many businesses across the globe with possibly classified information. Rachael Bale, a journalist at National Geographic, wrote that she found this kind of monitoring creepy. Another user, Kelly O’Mara, who is writer, stated that people really need to take into consideration how much information they have been feeding into Google Docs.

While hardly few will actually read it, the terms of service of Google Drive forbid spam, hate or violent speech, pornography, the spread of private information, etc. Only shared files are subjected to these rules, so the files that are kept private are apparently an exception.

A spokesman for Google stated that Docs and other files in Google Drive are regulated by an algorithm that keeps a watch on the abuse of its policies and if they are found to supposedly be a violation then it automatically blocks those files. Unlike a few systems, where the supposed matter is escalated to a moderator, here there is no human evaluation involved.

The spokesman also said that the primary reason its algorithm scans the files is for the prevention of computer viruses or spam from spreading. It is not clear if Google’s algorithm is utilised to keep a check on pornography or other types of misuse, or if the policy is equally applied to work as well as personal accounts.

After many users were locked out of their vital documents, the company stated that they made a code push that mistakenly marked a small percentage of Google Docs as abusive, which resulted in those documents getting blocked automatically. They also mentioned that a fix was already in place and all users will soon regain full access to their private docs. They issued an apology for the disruption while maintaining that their main focus is user safety.

While Google provides free of charge access to its “G Suite” of programs to its corporate users, Google Docs is available to consumers for free same as Gmail, Maps and its other services. Its company privacy policy mentions that it collects information from Google Docs along with other services, but it is not clarified whether this information is used for adverts.

A New Set of Eyes in Your Home: Cherry Home

There are a range of high security cameras out there, this one is a little different in monitoring your house. Cherry home made by Cherry labs, a startup company is the latest in AI technology. Besides this gadget being the eyes in your house it also senses your every movement and the layout of the room that it is in.

The cherry home cam can even identify you wherever you are in your home. The Cherry home sensors are able to detect your eye color, proportions of your body and even hair color. This gadget can do all this and more with its processor and sensors.

This gadget can also sense your laugh, scream and squeals and analyses them to see whether there is any danger. The info is then passed onto to an app that can be installed on any smartphone. The cherry home cam can also dial 911 in case of an emergency.

With the latest in AI technology, this gadget is able to triangulate the room it is in and give real time data to an app on your smartphone. So an individual can see all that is happening in the room that they are monitoring.

The gadget also notifies you in case there is an incident such as a person falling down. It is able to discern an individual’s laughter and scream when they are in danger.

Right now the Cherry home cam can do all these things but the startup company has also promised more in the future. According to the company, the Cherry cam will be able to sync in with your smart home and turn on lights when you need them. Another feature of cherry cam that is in the pipeline could be a breakthrough in what a security cam can do. The Cherry labs is reportedly in talks with pharmaceutical companies in trying to get the Cherry cam to identify when a person is in the process of getting a stroke and then to dial 911 or emergency services.

The gadget is packed with a binocular camera, infrared sensor for night time monitoring, an altimeter, a compass and an accelerometer. So how does this thing actually work? The cam sends data to a central processor with a high end PC to analyze all the data. It connects to other cameras in the house through a mesh WiFi network. It only connects to the cloud when it needs to transfer data to an app on your smartphone.

This Gadget also comes with a coin sized battery that is recharged by the processor cherry cam houses. There are also two batteries that are spares in the camera.

Now for the most important question at what price can a consumer get the incredible Cherry cam? The prices of the Cherry cam are not so conducive to buying them. A starter pack is available for 900$, a camera in this range would include 2 sensors and a processor while a pack of 4 sensors would cost you 1200$.