SciTech Bulletin 2.8
A special “Week of the Web” edition of your fortnightly dose of everything science and technology: Volume 2 Issue 8
In this edition of the SciTech Bulletin, we seek to highlight the latest inventions and discoveries in the fields of Computer Science and Communication technology, on the occasion of the International Internet Day observed on October 29th, which marked the beginning of Pragyan’s Week of the Web.
Learn more about Pragyan’s Week of the Web at the official portal.
Shelley: The bot that narrates macabre tales
A research team at MIT has released a chatbot called Shelley — named after Mary Shelley, the author of “Frankenstein”- that is capable of generating horror stories.
Shelley is a deep-learning powered AI system that is a combination of a learning algorithm and a recurrent neural network capable of learning from feedback. Trained with a huge dataset of over 140,000 stories contributed by amateur horror fiction writers, this bot is well trained to come up with wacky, unpredictable tales that test the limits of machine learning.
The bot is currently active on Twitter as @shelley_ai, where it tweets parts of a story with a #yourturn at the end. A human Twitter user can collaborate with it by tweeting back the continuation, to which Shelley will respond. This collaboration between man and machine will witness creativity and intelligence go hand in hand.
Secure WiFi: A Thing of the Past?
WPA2 (WiFi Protected Access) protocol has been the industry standard for network security for about 13 years now. This encryption methodology became widely adopted due to its high security features and compatibility across a wide range of hardware. Recently, however, a couple of researchers found a vulnerability in the encryption and were able to crack it. This method of attack is rightfully titled “KRACK”, and stands for Key Reinstallation Attack.
In most cases, the credentials of the client and the access points are verified by using special ‘handshake’ messages. KRACK exposes a vulnerability in this ‘handshake’ process and is able to manipulate and replay those messages. This tricks the devices into establishing insecure connections and hence, jeopardizing user data.
The vulnerability is an inherent flaw of the protocol itself, and is not device/implementation specific. Simply put, if the device in question is WiFi enabled, it is safe to assume that its security has been breached.
Ransomware breach wrecks havoc in Europe
A Ransomware named Bad Rabbit has created turmoil across Europe by demanding users a payment in bitcoins to grant access to the system. The malware, which has predominantly spread in Russia, Ukraine and Turkey, has its roots on the lines of WannaCry and Petya malwares responsible for similar outbreaks which occurred earlier this year. Initial reports also classified Bad rabbit as a variant under the Petyaware family.
With over 200 major companies affected, Bad Rabbit primarily works by ransoming 0.05 bitcoins valued at 285$ or Rs.18,480.Russian news agency Interfax and Fontanka were two major enterprises to be affected by this malware. In Ukraine, the Kiev Metro, Odessa International Airport and the Ministry of Infrastructure of Ukraine also fell prey to the attack.
Kaspersky labs which analysed the threat reported that the ransomware was downloaded as fake Adobe Flash players updates to lure victims into installing the malware unwittingly.
CERT-In The Indian Computer Emergency Response Team was quick to act in identifying the threat and has also issued a medium severity threat warning against the Bad Rabbit Ransomware. A general statement regarding Cyber protection and safety was also released.
Read the detailed article about the Bad Rabbit on The Hacker News to know more.
Decoding the mind using AI
In a bid to unravel the intricacies of the human mind, researchers at Purdue University have used Artificial Intelligence techniques to decode what the human brain sees. The process, that uses an algorithm called a convolutional neural network, interprets fMRI (functional magnetic resonance imaging) scans of people watching different videos, simulating a sort of mind-reading technology.
The researchers collected fMRI data from subjects watching video clips, which was then used to train the convolutional neural network model to predict the activity in the brain’s visual cortex. The model was used to decode fMRI data from the subjects to reconstruct the videos. It was able to accurately decode the data into specific image categories and correctly interpret what the person’s brain saw while watching the video.
This technology, apart from its applications in the field of Neuroscience, also boosts efforts to improve research in AI. Both these fields are highly interlinked. As efforts are being made to advance AI using brain-inspired concepts, we can also use AI to gain a deeper understanding of the functioning of the human brain.
Read more about this technology at ScienceDaily’s release.