System Bits: Dec. 6

Teaching computers to read
A multidisciplinary team of UCLA researchers has built a computational model that reflects how humans think and communicate, by designing an algorithm that examined nearly two million posts from popular parenting websites, thereby teaching computers to understand structured narratives within the flow of posts on the internet.

Managing large-scale data in this way highlights the overarching potential of machine learning, and demonstrates the capability to introduce counter-narratives into internet interactions, break up echo chambers and one day potentially help root out fact from fiction for social media users, the researchers noted.

UCLA researchers say machines could be taught to decipher false narratives as they proliferate.
(Source: UCLA)

Timothy Tangherlini, lead author and a self-described “computational folklorist” who teaches folklore, literature and cultural studies in the Scandinavian section of the UCLA College, said, “Our question was, could we devise computational methods to discover an emerging narrative framework underlying internet conversations that was possibly influencing the decision making of many people throughout the country or possibly world?”

Specifically, the researchers used sophisticated language modeling to review 1.99 million posts from two parenting sites with active user forums. They examined posts on Mothering.com — a site known to be a hub of anti-vaccine sentiment — and another parenting site (unnamed due to site privacy rules) where opinions on vaccinations were more varied. Those posts came from 40,056 users and were viewed 20.12 million times over a period of nearly nine years ending in 2012. Most users on both sites identified themselves as a mother.

The project was partially funded by a grant from the National Institutes of Health. Collaborating with Tangherlini were machine-learning expert Vwani Roychowdhury, UCLA professor of electrical engineering, and Dr. Roshan Bastani, a professor of health policy and management in the Fielding School of Public Health, and director of the UCLA Center for Prevention Research.

For the study, and based on his past scholarship in Danish folklore, Tangherlini and his colleagues came up with a broadly defined model of narrative, making that model a key part of the computational framework. They aligned this narrative model with nearly two million pieces of aggregated content from the parenting sites and, using natural language processing methods, were able to identify characters and the relationships between those characters, discovering the core of the underlying narratives.

On the basis of this work, they discovered that a large number of parents were not only going online to talk about vaccines, their distrust of institutions requiring them, or the perceived health risks of vaccinations, but also to seek out ways to acquire vaccination exemptions for their children.

Stories often emerge through conversation, the researchers noted, and the framework of the underlying narrative emerges through time as more and more stories are circulated, negotiated, aligned and reconfigured.

And while this study specifically applied to parents’ discussions about vaccination, the methods could be applied to any topic, said the researchers, who are pursuing follow up projects like incorporating a sequencing mechanism, which would track story plot.

Interestingly, the team said the way we learn about how stories take shape around any given topic can be applied to targeted messaging like advertising or fighting misinformation by allowing machine learning to automatically decipher false narratives as they proliferate.

For example, users exposed to particular anti-vaccination narrative could be presented with alternate narratives, based on well-tested public health paradigms, using the same extensive online advertising infrastructure currently used by the likes of Google, Facebook and Amazon.

Design your own custom drone
This fall’s new Federal Aviation Administration regulations have made drone flight easier than ever for both companies and consumers, and a new system from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) is the first to allow users to design, simulate, and build their own custom drone. Users can change the size, shape, and structure of their drone based on the specific needs they have for payload, cost, flight time, battery usage, and other factors.

A four-rotor “bunnycopter” developed at MIT’s Computer Science and Artificial Intelligence Laboratory features propellers at different heights. 
(Source: MIT CSAIL)



To demonstrate, researchers created a range of unusual-looking drones, including a five-rotor “pentacopter” and a rabbit-shaped “bunnycopter” with propellers of different sizes and rotors of different heights.

The interface lets users design drones with different propellers, rotors, and rods. It also provides guarantees that the drones it fabricates can take off, hover and land — which is no simple task considering the intricate technical trade-offs associated with drone weight, shape, and control.

The team reminded that today’s commercial drones only come in a small range of options, typically with an even number of rotors and upward-facing propellers. But there are many emerging use cases for other kinds of drones. For example, having an odd number of rotors might create a clearer view for a drone’s camera, or allow the drone to carry objects with unusual shapes. Further, designing less-conventional drones often requires expertise in multiple disciplines, including control systems, fabrication, and electronics.

The CSAIL group’s new system makes the process much easier. Users design drones by choosing from a database of parts and specifying their needs for things like payload, cost, and battery usage. The system computes the sizes of design elements like rod lengths and motor angles, and looks at metrics such as torque and thrust to determine whether the design will actually work. It also uses an “LQR controller” that takes information about a drone’s characteristics and surroundings to optimize its flight plan.

Prototype chip —- three atoms thick
Stanford University researchers reminded that for more than 50 years, chipmakers have invented ways to switch electricity on and off, generating the digital ones and zeroes that encode words, pictures, movies and other forms of data. But as researchers think about electronics for the next 50 years, they’ve begun to look beyond silicon to new types of materials that occur in single layers only three atoms thick – far thinner than modern silicon chips – yet are able to control electricity more efficiently to create those digital ones and zeroes. To his end, a team led by Stanford electrical engineering Associate Professor Eric Pop has demonstrated how it might be possible to mass-produce such atomically thin materials and electronics.

This would be useful because such thin materials would be transparent and flexible as well, in ways that would enable electronic devices that wouldn’t be possible to make with silicon.

The researchers etched a nanoscale image of the Stanford tree onto an ultrathin chip, using the same technique that could one day create electronic circuits. (Source: Associate Professor Eric Pop’s Lab, Stanford University)

The goal of the research was to develop a manufacturing process to turn single-layer chips into practical realities. The first atomically thin material was measured in 2004 when scientists observed that graphene – a material related to the “lead” in pencils – could be isolated in layers the thickness of a single carbon atom, and the scientists who made this finding shared the 2010 Nobel Prize in Physics.

It is on the issue of manufacturability where the Stanford team members made a big advance. They started with a single layer of material called molybdenum disulfide. The name describes its sandwich-like structure: a sheet of molybdenum atoms between two layers of sulfur. Previous research had shown that molybdenum disulfide made a good switch, controlling electricity to create digital ones and zeroes.

The question was whether the team could manufacture a molybdenum disulfide crystal big enough to form a chip. That requires building a crystal roughly the size of your thumbnail. This may not sound like a big deal until you consider the aspect ratio of the crystal required: a chip just three atoms thick but the size of your thumbnail is like a single sheet of paper big enough to cover the entire Stanford campus.

The Stanford team manufactured that sheet by depositing three layers of atoms into a crystalline structure 25 million times wider than it is thick. Smithe achieved this by making ingenious refinements to a manufacturing process called chemical vapor deposition. This approach essentially incinerates small amounts of sulfur and molybdenum until the atoms vaporize like soot. The atoms then deposit as an ultra-thin crystalline layer on a “handle” substrate, which can be glass or even silicon.

However, the researchers said their job was not done; they still had to pattern the material into electrical switches and to understand their operation. For this, they made use of a recent advance led by English, who discovered that extremely clean deposition conditions are essential to form good metallic contacts with the molybdenum disulfide layers. The wealth of new experimental data available now in the lab has also enabled Suryavanshi to craft accurate computer models of the new materials and to begin predicting their collective behavior as circuit components.

Related Stories
Teaching Computers To See
First of two parts: Vision processing is a specialized design task, requiring dedicated circuitry to achieve low power and latency.
Making Drones Secure
Current-generation drones are imperiled by multiple security weaknesses. Is the semiconductor industry doing enough to address the problem?
System Research Bits: Nov. 29
300mm quantum demo; photonic crystals; IoT device biosignal measuring.