Daily Archives: January 20, 2018

Active learning machine learns to create new quantum experiments

Posted on by

We present an autonomous learning model which learns to design such complex experiments, without relying on previous knowledge or often flawed intuition. Our system not only learns how to design desired experiments more efficiently than the best previous approaches, but in the process also discovers nontrivial experimental techniques. Our work demonstrates that learning machines can offer dramatic advances in how experiments are generated.
[…]
The artificial intelligence system learns to create a variety of entangled states and improves the efficiency of their realization. In the process, the system autonomously (re)discovers experimental techniques which are only now becoming standard in modern quantum optical experiments—a trait which was not explicitly demanded from the system but emerged through the process of learning. Such features highlight the possibility that machines could have a significantly more creative role in future research.

Source: Active learning machine learns to create new quantum experiments

The artificial agent develops new experiments by virtually placing mirrors, prisms or beam splitters on a virtual lab table. If its actions lead to a meaningful result, the agent has a higher chance of finding a similar sequence of actions in the future. This is known as a reinforcement learning strategy.

Read more at: https://phys.org/news/2018-01-artificial-agent-quantum.html#jCp

Breakthrough study shows how plants sense the world

Posted on by

Plants lack eyes and ears, but they can still see, hear, smell and respond to environmental cues and dangers—especially to virulent pathogens. They do this with the aid of hundreds of membrane proteins that can sense microbes or other stresses.

Only a small portion of these sensing proteins have been studied through classical genetics, and knowledge on how these sensors function by forming complexes with one another is scarce. Now, an international team of researchers from four nations—including Shahid Mukhtar, Ph.D., and graduate student Timothy “TC” Howton at the University of Alabama at Birmingham—has created the first network map for 200 of these proteins. The map shows how a few key proteins act as master nodes critical for network integrity, and the map also reveals unknown interactions.
[…]
The model plant Arabidopsis thaliana contains more than 600 different receptor kinases—50 times more than humans—that are critical for plant growth, development, immunity and stress response. Until now, only a handful had known functions, and little was known about how the receptors might interact with each to coordinate responses to often-conflicting signals.

For the Nature study, the Belkhadir lab tested interactions between extracellular domains of the receptors in a pairwise manner, working with more than 400 extracellular domains of the LRR-receptor kinases and performing 40,000 interaction tests.

Positive interactions were used to produce an interaction map displaying how those receptor kinases interact with one another, in a total of 567 high-confidence interactions.
[…]
At UAB, Mukhtar and Howton tested 372 intracellular domains of the LRR-receptor kinases whose extracellular domains had shown high-confidence interactions, to see if the intracellular domains also showed strong interactions. More than half did, suggesting that the formation of these receptor complexes is required for signal perception and downstream signal transduction. This also indicates a validation of the biological significance of the extracellular domain interaction
[…]
The Nature study included two major surprises, says Adam Mott, Ph.D., University of Toronto. LRR-receptor kinases that have small extracellular domains interacted with other LRR-receptor kinases more often than those that have large domains. This suggests that the small receptor kinases evolved to coordinate actions of the other receptors. Second, researchers identified several unknown LRR-receptor kinases that appear critical for network integrity.

Source: Breakthrough study shows how plants sense the world

So yes, vegetarians, plants do live and feel and see and detect, you murderers!

American Reich restarts dodgy spying program – just as classified surveillance abuse memo emerges

Posted on by

The US Senate reauthorized a controversial NSA spying program on Thursday – and then, because it’s 2018 and nothing matters any more, embarked on a partisan battle over a confidential memo that outlines Uncle Sam’s alleged abuse of surveillance powers.

Despite numerous appeals, press conferences, competing legislation and speeches outlining abuse of the program, on Thursday a majority of senators ignored pleas for a proper warrant requirement to be added to the program – that would require the Feds to always go to a judge before searching the communications of a US citizen – and voted to continue the surveillance for a further six years.
[…]

However, the agents won’t need a warrant if they are looking into…

Death, kidnapping, serious bodily injury, offense against a minor, destruction of critical infrastructure, cybersecurity, transnational crime, and human trafficking

…which are basically the crimes the FBI investigates. Ergo, it’s unlikely the Feds will seek warrants to search the NSA’s section 702 data stores for stuff on American citizens.

Just hours after the section 702 program was given the final green light before the president can sign on the dotted line, the Senate’s intelligence committee approved the release of a confidential four-page memo alleging previous abuse of the FISA spying program to the rest of Congress. The public is unable to see it.

The mysterious missive was drafted by House intelligence committee chairman Devin Nunes (R-CA), and of course it could be looney-tunes nonsense. Regardless, a number of lawmakers who only now just read the memo have said that had they been aware of the misconduct detailed in the memo, they would not having voted for the reauthorization of section 702 of the FISA Amendments Act.

Republican lawmakers in particular, having seen the report, embarked on a fiercely partisan campaign accusing the Obama administration of snooping on the Trump presidential campaign using the foreigner-targeting FISA laws.
[…]
The hypocrisy is stunning, even for Congress. One moment, Republicans insist a Big Brother program is needed to foil terrorists abroad, ignoring its ability to pry into the lives of Americans. The next moment, Republicans are upset the same set of laws were indeed used to pry into the lives of Americans – some of the folks working for Team Trump.
[…]
Congresscriters who now claim to be shocked – shocked! – about FISA’s sweeping capabilities – have been willfully ignoring determined efforts in both the House and the Senate in recent weeks to have a full debate about the extent of spying powers that the US government possesses
[…]

In one part of that speech, he even went into great detail over how the Director of National Intelligence had publicly denied that Uncle Sam was able to intercept communications between US citizens on US soil – and then, when challenged subsequently, claimed to have heard a different question.

When Wyden asked the same question again, the director refused to answer, claiming that it was classified. “How can a topic in which the director of national intelligence has already given an answer in public suddenly become classified?” asked Wyden in his speech.

But if all that wasn’t enough, we will all likely be subject to one more head-holding display of hypocrisy when President Trump signs the reauthorization bill into law – despite the fact congressfolk are railing against the same set of FISA laws being used to spy on his campaign.

Source: America restarts dodgy spying program – just as classified surveillance abuse memo emerges • The Register

Security Breaches Don’t Affect Stock Price. Or don’t they?

Posted on by

Abstract: This report assesses the impact disclosure of data breaches has on the total returns and volatility of the affected companies’ stock, with a focus on the results relative to the performance of the firms’ peer industries, as represented through selected indices rather than the market as a whole. Financial performance is considered over a range of dates from 3 days post-breach through 6 months post-breach, in order to provide a longer-term perspective on the impact of the breach announcement.

Key findings:

While the difference in stock price between the sampled breached companies and their peers was negative (1.13%) in the first 3 days following announcement of a breach, by the 14th day the return difference had rebounded to + 0.05%, and on average remained positive through the period assessed.

For the differences in the breached companies’ betas and the beta of their peer sets, the differences in the means of 8 months pre-breach versus post-breach was not meaningful at 90, 180, and 360 day post-breach periods.

For the differences in the breached companies’ beta correlations against the peer indices pre- and post-breach, the difference in the means of the rolling 60 day correlation 8 months pre- breach versus post-breach was not meaningful at 90, 180, and 360 day post-breach periods.

In regression analysis, use of the number of accessed records, date, data sensitivity, and malicious versus accidental leak as variables failed to yield an R2 greater than 16.15% for response variables of 3, 14, 60, and 90 day return differential, excess beta differential, and rolling beta correlation differential, indicating that the financial impact on breached companies was highly idiosyncratic.

Based on returns, the most impacted industries at the 3 day post-breach date were U.S. Financial Services, Transportation, and Global Telecom. At the 90 day post-breach date, the three most impacted industries were U.S. Financial Services, U.S. Healthcare, and Global Telecom.

The market isn’t going to fix this. If we want better security, we need to regulate the market.

Source: Security Breaches Don’t Affect Stock Price – Schneier on Security

However, the dataset:

The analysis began with a dataset of 235 recorded data breaches dating back to 2005

is very very small and misses some of the huge breaches such as Equifax.
There is a very telling table in the results that does show that if a breach is hugely public, then share prices do indeed plummet:

So it may also have something to do with how the company handles the breach and how much media attention is out there.