Tag Archives: Massachusetts Institute of Technology. MIT

AI (artificial intelligence) for Good Global Summit from May 15 – 17, 2018 in Geneva, Switzerland: details and an interview with Frederic Werner

With all the talk about artificial intelligence (AI), a lot more attention seems to be paid to apocalyptic scenarios: loss of jobs, financial hardship, loss of personal agency and privacy, and more with all of these impacts being described as global. Still, there are some folks who are considering and working on ‘AI for good’.

If you’d asked me, the International Telecommunications Union (ITU) would not have been my first guess (my choice would have been United Nations Educational, Scientific and Cultural Organization [UNESCO]) as an agency likely to host the 2018 AI for Good Global Summit. But, it turns out the ITU is a UN (United Nations agency) and, according to its Wikipedia entry, it’s an intergovernmental public-private partnership, which may explain the nature of the participants in the upcoming summit.

The news

First, there’s a May 4, 2018 ITU media advisory (received via email or you can find the full media advisory here) about the upcoming summit,

Artificial Intelligence (AI) is now widely identified as being able to address the greatest challenges facing humanity – supporting innovation in fields ranging from crisis management and healthcare to smart cities and communications networking.

The second annual ‘AI for Good Global Summit’ will take place 15-17 May [2018] in Geneva, and seeks to leverage AI to accelerate progress towards the United Nations’ Sustainable Development Goals and ultimately benefit humanity.

WHAT: Global event to advance ‘AI for Good’ with the participation of internationally recognized AI experts. The programme will include interactive high-level panels, while ‘AI Breakthrough Teams’ will propose AI strategies able to create impact in the near term, guided by an expert audience of mentors representing government, industry, academia and civil society – through interactive sessions. The summit will connect AI innovators with public and private-sector decision-makers, building collaboration to take promising strategies forward.

A special demo & exhibit track will feature innovative applications of AI designed to: protect women from sexual violence, avoid infant crib deaths, end child abuse, predict oral cancer, and improve mental health treatments for depression – as well as interactive robots including: Alice, a Dutch invention designed to support the aged; iCub, an open-source robot; and Sophia, the humanoid AI robot.

WHEN: 15-17 May 2018, beginning daily at 9 AM

WHERE: ITU Headquarters, 2 Rue de Varembé, Geneva, Switzerland (Please note: entrance to ITU is now limited for all visitors to the Montbrillant building entrance only on rue Varembé).

WHO: Confirmed participants to date include expert representatives from: Association for Computing Machinery, Bill and Melinda Gates Foundation, Cambridge University, Carnegie Mellon, Chan Zuckerberg Initiative, Consumer Trade Association, Facebook, Fraunhofer, Google, Harvard University, IBM Watson, IEEE, Intellectual Ventures, ITU, Microsoft, Massachusetts Institute of Technology (MIT), Partnership on AI, Planet Labs, Shenzhen Open Innovation Lab, University of California at Berkeley, University of Tokyo, XPRIZE Foundation, Yale University – and the participation of “Sophia” the humanoid robot and “iCub” the EU open source robotcub.

The interview

Frederic Werner, Senior Communications Officer at the International Telecommunication Union and and one of the organizers of the AI for Good Global Summit 2018 kindly took the time to speak to me and provide a few more details about the upcoming event.

Werner noted that the 2018 event grew out of a much smaller 2017 ‘workshop’ and first of its kind, about beneficial AI which this year has ballooned in size to 91 countries (about 15 participants are expected from Canada), 32 UN agencies, and substantive representation from the private sector. The 2017 event featured Dr. Yoshua Bengio of the University of Montreal  (Université de Montréal) was a featured speaker.

“This year, we’re focused on action-oriented projects that will help us reach our Sustainable Development Goals (SDGs) by 2030. We’re looking at near-term practical AI applications,” says Werner. “We’re matchmaking problem-owners and solution-owners.”

Academics, industry professionals, government officials, and representatives from UN agencies are gathering  to work on four tracks/themes:

In advance of this meeting, the group launched an AI repository (an action item from the 2017 meeting) on April 25, 2018 inviting people to list their AI projects (from the ITU’s April 25, 2018? AI repository news announcement),

ITU has just launched an AI Repository where anyone working in the field of artificial intelligence (AI) can contribute key information about how to leverage AI to help solve humanity’s greatest challenges.

This is the only global repository that identifies AI-related projects, research initiatives, think-tanks and organizations that aim to accelerate progress on the 17 United Nations’ Sustainable Development Goals (SDGs).

To submit a project, just press ‘Submit’ on the AI Repository site and fill in the online questionnaire, providing all relevant details of your project. You will also be asked to map your project to the relevant World Summit on the Information Society (WSIS) action lines and the SDGs. Approved projects will be officially registered in the repository database.

Benefits of participation on the AI Repository include:

WSIS Prizes recognize individuals, governments, civil society, local, regional and international agencies, research institutions and private-sector companies for outstanding success in implementing development oriented strategies that leverage the power of AI and ICTs.

Creating the AI Repository was one of the action items of last year’s AI for Good Global Summit.

We are looking forward to your submissions.

If you have any questions, please send an email to: ai@itu.int

“Your project won’t be visible immediately as we have to vet the submissions to weed out spam-type material and projects that are not in line with our goals,” says Werner. That said, there are already 29 projects in the repository. As you might expect, the UK, China, and US are in the repository but also represented are Egypt, Uganda, Belarus, Serbia, Peru, Italy, and other countries not commonly cited when discussing AI research.

Werner also pointed out in response to my surprise over the ITU’s role with regard to this AI initiative that the ITU is the only UN agency which has 192* member states (countries), 150 universities, and over 700 industry members as well as other member entities, which gives them tremendous breadth of reach. As well, the organization, founded originally in 1865 as the International Telegraph Convention, has extensive experience with global standardization in the information technology and telecommunications industries. (See more in their Wikipedia entry.)

Finally

There is a bit more about the summit on the ITU’s AI for Good Global Summit 2018 webpage,

The 2nd edition of the AI for Good Global Summit will be organized by ITU in Geneva on 15-17 May 2018, in partnership with XPRIZE Foundation, the global leader in incentivized prize competitions, the Association for Computing Machinery (ACM) and sister United Nations agencies including UNESCO, UNICEF, UNCTAD, UNIDO, Global Pulse, UNICRI, UNODA, UNIDIR, UNODC, WFP, IFAD, UNAIDS, WIPO, ILO, UNITAR, UNOPS, OHCHR, UN UniversityWHO, UNEP, ICAO, UNDP, The World Bank, UN DESA, CTBTOUNISDRUNOG, UNOOSAUNFPAUNECE, UNDPA, and UNHCR.

The AI for Good series is the leading United Nations platform for dialogue on AI. The action​​-oriented 2018 summit will identify practical applications of AI and supporting strategies to improve the quality and sustainability of life on our planet. The summit will continue to formulate strategies to ensure trusted, safe and inclusive development of AI technologies and equitable access to their benefits.

While the 2017 summit sparked the first ever inclusive global dialogue on beneficial AI, the action-oriented 2018 summit will focus on impactful AI solutions able to yield long-term benefits and help achieve the Sustainable Development Goals. ‘Breakthrough teams’ will demonstrate the potential of AI to map poverty and aid with natural disasters using satellite imagery, how AI could assist the delivery of citizen-centric services in smart cities, and new opportunities for AI to help achieve Universal Health Coverage, and finally to help achieve transparency and explainability in AI algorithms.

Teams will propose impactful AI strategies able to be enacted in the near term, guided by an expert audience of mentors representing government, industry, academia and civil society. Strategies will be evaluated by the mentors according to their feasibility and scalability, potential to address truly global challenges, degree of supporting advocacy, and applicability to market failures beyond the scope of government and industry. The exercise will connect AI innovators with public and private-sector decision-makers, building collaboration to take promising strategies forward.

“As the UN specialized agency for information and communication technologies, ITU is well placed to guide AI innovation towards the achievement of the UN Sustainable Development ​Goals. We are providing a neutral close quotation markplatform for international dialogue aimed at ​building a ​common understanding of the capabilities of emerging AI technologies.​​” Houlin Zhao, Secretary General ​of ITU​

Should you be close to Geneva, it seems that registration is still open. Just go to the ITU’s AI for Good Global Summit 2018 webpage, scroll the page down to ‘Documentation’ and you will find a link to the invitation and a link to online registration. Participation is free but I expect that you are responsible for your travel and accommodation costs.

For anyone unable to attend in person, the summit will be livestreamed (webcast in real time) and you can watch the sessions by following the link below,

https://www.itu.int/en/ITU-T/AI/2018/Pages/webcast.aspx

For those of us on the West Coast of Canada and other parts distant to Geneva, you will want to take the nine hour difference between Geneva (Switzerland) and here into account when viewing the proceedings. If you can’t manage the time difference, the sessions are being recorded and will be posted at a later date.

*’132 member states’ corrected to ‘192 member states’ on May 11, 2018 at 1500 hours PDT.

Understanding how carbon nanotubes grow and self-organize is key to better production

This research may help to commercialize use of carbon nanotubes (CNTs), a  ‘magical’ nanoscale material with great promise and great difficulties (standardizing production being one of the main difficulties). A Feb. 10, 2017 news item on phys.org describes how researchers at the Lawrence Livermore National Laboratory (LLNL) and other collaborators have recorded carbon nanotubes self-organizing,

For the first time, Lawrence Livermore National Laboratory scientists and collaborators have captured a movie of how large populations of carbon nanotubes grow and align themselves.

Understanding how carbon nanotubes (CNT) nucleate, grow and self-organize to form macroscale materials is critical for application-oriented design of next-generation supercapacitors, electronic interconnects, separation membranes and advanced yarns and fabrics.

A Feb. 9, 2017 LLNL news release, which originated the news item, provides more information about the research (Note: Links have been removed),

New research by LLNL scientist Eric Meshot and colleagues from Brookhaven National Laboratory (link is external) (BNL) and Massachusetts Institute of Technology (link is external) (MIT) has demonstrated direct visualization of collective nucleation and self-organization of aligned carbon nanotube films inside of an environmental transmission electron microscope (ETEM).

In a pair of studies reported in recent issues of Chemistry of Materials (link is external) and ACS Nano (link is external), the researchers leveraged a state-of-the-art kilohertz camera in an aberration-correction ETEM at BNL to capture the inherently rapid processes that govern the growth of these exciting nanostructures.

Among other phenomena discovered, the researchers are the first to provide direct proof of how mechanical competition among neighboring carbon nanotubes can simultaneously promote self-alignment while also frustrating and limiting growth.

“This knowledge may enable new pathways toward mitigating self-termination and promoting growth of ultra-dense and aligned carbon nanotube materials, which would directly impact several application spaces, some of which are being pursued here at the Laboratory,” Meshot said.

Meshot has led the CNT synthesis development at LLNL for several projects, including those supported by the Laboratory Directed Research and Development (LDRD) program and the Defense Threat Reduction Agency (link is external) (DTRA) that use CNTs as fluidic nanochannels for applications ranging from single-molecule detection to macroscale membranes for breathable and protective garments.

Here’s a link to and a citation for the both of the papers mentioned in the news release,

Measurement of the Dewetting, Nucleation, and Deactivation Kinetics of Carbon Nanotube Population Growth by Environmental Transmission Electron Microscopy by Mostafa Bedewy, B. Viswanath, Eric R. Meshot, Dmitri N. Zakharov, Eric A. Stach, and A. John Hart. Chem. Mater., 2016, 28 (11), pp 3804–3813 DOI: 10.1021/acs.chemmater.6b00798 Publication Date (Web): May 23, 2016

Copyright © 2016 American Chemical Society

Real-Time Imaging of Self-Organization and Mechanical Competition in Carbon Nanotube Forest Growth by Viswanath Balakrishnan, Mostafa Bedewy, Eric R. Meshot, Sebastian W. Pattinson, Erik S. Polsen, Fabrice Laye, Dmitri N. Zakharov, Eric A. Stach, and A. John Hart. ACS Nano, 2016, 10 (12), pp 11496–11504 DOI: 10.1021/acsnano.6b07251 Publication Date (Web): November 23, 2016

Copyright © 2016 American Chemical Society

Both papers are behind a paywall.

The researchers have also provided this image which allows you to appreciate the difference between a ‘scientific’ version of the work and an artistic version,

This transmission electron microscope image shows growth of a dense carbon nanotube population. Courtesy: LLNL

“Breaking Me Softly” at the nanoscale

“Breaking Me Softly” sounds like a song title but in this case the phrase as been coined to describe a new technique for controlling materials at the nanoscale according to a June 6, 2016 news item on ScienceDaily,

A finding by a University of Central Florida researcher that unlocks a means of controlling materials at the nanoscale and opens the door to a new generation of manufacturing is featured online in the journal Nature.

Using a pair of pliers in each hand and gradually pulling taut a piece of glass fiber coated in plastic, associate professor Ayman Abouraddy found that something unexpected and never before documented occurred — the inner fiber fragmented in an orderly fashion.

“What we expected to see happen is NOT what happened,” he said. “While we thought the core material would snap into two large pieces, instead it broke into many equal-sized pieces.”

He referred to the technique in the Nature article title as “Breaking Me Softly.”

A June 6, 2016 University of Central Florida (UCF) news release (also on EurekAlert) by Barbara Abney, which originated the news item, expands on the theme,

The process of pulling fibers to force the realignment of the molecules that hold them together, known as cold drawing, has been the standard for mass production of flexible fibers like plastic and nylon for most of the last century.

Abouraddy and his team have shown that the process may also be applicable to multi-layered materials, a finding that could lead to the manufacturing of a new generation of materials with futuristic attributes.

“Advanced fibers are going to be pursuing the limits of anything a single material can endure today,” Abouraddy said.

For example, packaging together materials with optical and mechanical properties along with sensors that could monitor such vital sign as blood pressure and heart rate would make it possible to make clothing capable of transmitting vital data to a doctor’s office via the Internet.

The ability to control breakage in a material is critical to developing computerized processes for potential manufacturing, said Yuanli Bai, a fracture mechanics specialist in UCF’s College of Engineering and Computer Science.

Abouraddy contacted Bai, who is a co-author on the paper, about three years ago and asked him to analyze the test results on a wide variety of materials, including silicon, silk, gold and even ice.

He also contacted Robert S. Hoy, a University of South Florida physicist who specializes in the properties of materials like glass and plastic, for a better understanding of what he found.

Hoy said he had never seen the phenomena Abouraddy was describing, but that it made great sense in retrospect.

The research takes what has traditionally been a problem in materials manufacturing and turned it into an asset, Hoy said.

“Dr. Abouraddy has found a new application of necking” –  a process that occurs when cold drawing causes non-uniform strain in a material, Hoy said.  “Usually you try to prevent necking, but he exploited it to do something potentially groundbreaking.”

The necking phenomenon was discovered decades ago at DuPont and ushered in the age of textiles and garments made of synthetic fibers.

Abouraddy said that cold-drawing is what makes synthetic fibers like nylon and polyester useful. While those fibers are initially brittle, once cold-drawn, the fibers toughen up and become useful in everyday commodities. This discovery at DuPont at the end of the 1920s ushered in the age of textiles and garments made of synthetic fibers.

Only recently have fibers made of multiple materials become possible, he said.  That research will be the centerpiece of a $317 Million U.S. Department of Defense program focused on smart fibers that Abouraddy and UCF will assist with.   The Revolutionary Fibers and Textiles Manufacturing Innovation Institute (RFT-MII), led by the Massachusetts Institute of Technology, will incorporate research findings published in the Nature paper, Abouraddy said.

The implications for manufacturing of the smart materials of the future are vast.

By controlling the mechanical force used to pull the fiber and therefore controlling the breakage patterns, materials can be developed with customized properties allowing them to interact with each other and eternal forces such as the sun (for harvesting energy) and the internet in customizable ways.

A co-author on the paper, Ali P. Gordon, an associate professor in the Department of Mechanical & Aerospace Engineering and director of UCF’s Mechanics of Materials Research Group said that the finding is significant because it shows that by carefully controlling the loading condition imparted to the fiber, materials can be developed with tailored performance attributes.

“Processing-structure-property relationships need to be strategically characterized for complex material systems. By combining experiments, microscopy, and computational mechanics, the physical mechanisms of the fragmentation process were more deeply understood,” Gordon said.

Abouraddy teamed up with seven UCF scientists from the College of Optics & Photonics and the College of Engineering & Computer Science (CECS) to write the paper.   Additional authors include one researcher each from the Massachusetts Institute of Technology, Nanyang Technological University in Singapore and the University of South Florida.

Here’s a link to and a citation for the paper,

Controlled fragmentation of multimaterial fibres and films via polymer cold-drawing by Soroush Shabahang, Guangming Tao, Joshua J. Kaufman, Yangyang Qiao, Lei Wei, Thomas Bouchenot, Ali P. Gordon, Yoel Fink, Yuanli Bai, Robert S. Hoy & Ayman F. Abouraddy. Nature (2016) doi:10.1038/nature17980 Published online  06 June 2016

This paper is behind a paywall.

Fermionic atoms and the microscopes that can see them

The new fermionic microscope built at the Massachusetts Institute of Technology (MIT) allows you to image 1000 or more fermionic atoms according to a May 13, 2015 news item on ScienceDaily,

Fermions are the building blocks of matter, interacting in a multitude of permutations to give rise to the elements of the periodic table. Without fermions, the physical world would not exist.

Examples of fermions are electrons, protons, neutrons, quarks, and atoms consisting of an odd number of these elementary particles. Because of their fermionic nature, electrons and nuclear matter are difficult to understand theoretically, so researchers are trying to use ultracold gases of fermionic atoms as stand-ins for other fermions.

But atoms are extremely sensitive to light: When a single photon hits an atom, it can knock the particle out of place — an effect that has made imaging individual fermionic atoms devilishly hard.

Now a team of MIT physicists has built a microscope that is able to see up to 1,000 individual fermionic atoms. The researchers devised a laser-based technique to trap and freeze fermions in place, and image the particles simultaneously.

A May 13, 2015 MIT news release, which originated the news item, provides intriguing detail about the microscope and fascinating insight into fermions (for those who are interested but not expert and sufficiently brave to endure certain failure to understand everything in this piece),

The new imaging technique uses two laser beams trained on a cloud of fermionic atoms in an optical lattice. The two beams, each of a different wavelength, cool the cloud, causing individual fermions to drop down an energy level, eventually bringing them to their lowest energy states — cool and stable enough to stay in place. At the same time, each fermion releases light, which is captured by the microscope and used to image the fermion’s exact position in the lattice — to an accuracy better than the wavelength of light.

With the new technique, the researchers are able to cool and image over 95 percent of the fermionic atoms making up a cloud of potassium gas. Martin Zwierlein, a professor of physics at MIT, says an intriguing result from the technique appears to be that it can keep fermions cold even after imaging.

“That means I know where they are, and I can maybe move them around with a little tweezer to any location, and arrange them in any pattern I’d like,” Zwierlein says.

Zwierlein and his colleagues, including first author and graduate student Lawrence Cheuk, have published their results today in the journal Physical Review Letters.

Seeing fermions from bosons

For the past two decades, experimental physicists have studied ultracold atomic gases of the two classes of particles: fermions and bosons — particles such as photons that, unlike fermions, can occupy the same quantum state in limitless numbers. In 2009, physicist Markus Greiner at Harvard University devised a microscope that successfully imaged individual bosons in a tightly spaced optical lattice. This milestone was followed, in 2010, by a second boson microscope, developed by Immanuel Bloch’s group at the Max Planck Institute of Quantum Optics.

These microscopes revealed, in unprecedented detail, the behavior of bosons under strong interactions. However, no one had yet developed a comparable microscope for fermionic atoms.

“We wanted to do what these groups had done for bosons, but for fermions,” Zwierlein says. “And it turned out it was much harder for fermions, because the atoms we use are not so easily cooled. So we had to find a new way to cool them while looking at them.”

Techniques to cool atoms ever closer to absolute zero have been devised in recent decades. Carl Wieman, Eric Cornell, and MIT’s Wolfgang Ketterle were able to achieve Bose-Einstein condensation in 1995, a milestone for which they were awarded the 2001 Nobel Prize in physics. Other techniques include a process using lasers to cool atoms from 300 degrees Celsius to a few ten-thousandths of a degree above absolute zero.

A clever cooling technique

And yet, to see individual fermionic atoms, the particles need to be cooled further still. To do this, Zwierlein’s group created an optical lattice using laser beams, forming a structure resembling an egg carton, each well of which could potentially trap a single fermion. Through various stages of laser cooling, magnetic trapping, and further evaporative cooling of the gas, the atoms were prepared at temperatures just above absolute zero — cold enough for individual fermions to settle onto the underlying optical lattice. The team placed the lattice a mere 7 microns from an imaging lens, through which they hoped to see individual fermions.

However, seeing fermions requires shining light on them, causing a photon to essentially knock a fermionic atom out of its well, and potentially out of the system entirely.

“We needed a clever technique to keep the atoms cool while looking at them,” Zwierlein says.

His team decided to use a two-laser approach to further cool the atoms; the technique manipulates an atom’s particular energy level, or vibrational energy. Each atom occupies a certain energy state — the higher that state, the more active the particle is. The team shone two laser beams of differing frequencies at the lattice. The difference in frequencies corresponded to the energy between a fermion’s energy levels. As a result, when both beams were directed at a fermion, the particle would absorb the smaller frequency, and emit a photon from the larger-frequency beam, in turn dropping one energy level to a cooler, more inert state. The lens above the lattice collects the emitted photon, recording its precise position, and that of the fermion.

Zwierlein says such high-resolution imaging of more than 1,000 fermionic atoms simultaneously would enhance our understanding of the behavior of other fermions in nature — particularly the behavior of electrons. This knowledge may one day advance our understanding of high-temperature superconductors, which enable lossless energy transport, as well as quantum systems such as solid-state systems or nuclear matter.

“The Fermi gas microscope, together with the ability to position atoms at will, might be an important step toward the realization of a quantum computer based on fermions,” Zwierlein says. “One would thus harness the power of the very same intricate quantum rules that so far hamper our understanding of electronic systems.”

Zwierlein says it is a good time for Fermi gas microscopists: Around the same time his group first reported its results, teams from Harvard and the University of Strathclyde in Glasgow also reported imaging individual fermionic atoms in optical lattices, indicating a promising future for such microscopes.

Zoran Hadzibabic, a professor of physics at Trinity College [University of Cambridge, UK], says the group’s microscope is able to detect individual atoms “with almost perfect fidelity.”

“They detect them reliably, and do so without affecting their positions — that’s all you want,” says Hadzibabic, who did not contribute to the research. “So far they demonstrated the technique, but we know from the experience with bosons that that’s the hardest step, and I expect the scientific results to start pouring out.”

Here’s a link to and a citation for the published paper,

Quantum-Gas Microscope for Fermionic Atoms by Lawrence W. Cheuk, Matthew A. Nichols, Melih Okan, Thomas Gersdorf, Vinay V. Ramasesh, Waseem S. Bakr, Thomas Lompe, and Martin W. Zwierlein. Phys. Rev. Lett. 114, 193001 – Published 13 May 2015 (print: Vol. 114, Iss. 19 — 15 May 2015) DOI: http://dx.doi.org/10.1103/PhysRevLett.114.193001

I believe this paper is behind a paywall.

There is an earlier version available on arXiv.org,

A Quantum Gas Microscope for Fermionic Atoms by Lawrence W. Cheuk, Matthew A. Nichols, Melih Okan, Thomas Gersdorf, Vinay V. Ramasesh, Waseem S. Bakr, Thomas Lompe, Martin W. Zwierlein. (Submitted on 9 Mar 2015 (v1), last revised 10 Mar 2015 (this version, v2))

This an open access website.