Vol.

IX
Issue- II
November 2020

E-Newsletter
COMPUTER SCIENCE & ENGINEERING

EDITORIAL BOARD Vision & Mission of Department

EDITOR
MS. SHRUTI AGARWAL
(Assistant Professor) VISION

DESIGNER Engineering the future of the nation by transforming the students to

KAPIL SAXENA
beskilled technocrats, innovative leaders and environmentally receptive
(Lab Instructor)
citizens. The Vision of the department is to carve the youth as dynamic,
STUDENTREPRESENTATIVES competent,valued and knowledgeable professionals who shall lead the
Nation to a better future.
NAME CLASS
MISSION
IRAM WARSI M.TECH -
2019
➢ To flourish the SRMS as the World Leader in Computer Science &
HARSIMAR SINGH CS1 - 2017
SETHI
Engineering through continuous research & development directed
RAYMAN GANGWAR CS2 - 2017 towards the betterment of the society.
CHARU AGARWAL CS1 - 2018 ➢ To establish the cooperative learning environment for facilitating the
HARSHITA MAURYA CS2 – 2018
quality academics, state-of-the art research and remarkable
BHAVISHYA GUPTA CS1 – 2019
TAPOSH CHANDRA CS2 - 2019 development activities.
SHARMA
➢ To establish World Class resources especially Research &
CONTENTS
Development Laboratories, Value Addition courses etc. for the in-
Spiritual Corner 3-3 house up gradation & community services.
Latest Updates World 3-5 ➢ To groom the students into Industry – Ready Professionals through
Engineering &Technology
Student Corner 5-7 arigorous training in a self- disciplined environment.

Faculty Arena 7-12 ➢ To groom the learned pool of faculty in accordance with the recent
advancements in the field of Computer Science & Engineering.

Department of Computer Science & Engineering

SHRI RAM MURTI SMARAK COLLEGE OF ENGINEERING& TECHNOLOGY, BAREILLY
Ram Murti Puram, 13 Km, Bareilly- Nainital
1 Road, Bhojipura, Bareilly-243202, UP, India.
Phone: 91-581-2582246, 2582249, 2582331, FAX: 91-581-2582330
E-mail: csenewsletter@srmscet.edu, Website: http://www.srms.ac.in/cet
―Educationisthepassporttothefuture,fortomorrowbelongstothosewhoprepare for it
today.‖
-Malcolm X

―Bemiserable.Ormotivateyourself.Whateverhastobedone,it'salwaysyour choice.‖
-Wayne Dyer

―Youwilleitherstepforwardintogrowth,oryouwillstepbackwardintosafety.‖
-Abraham Maslow

―Ifamanemptieshispurseintohishead,nomancantakeitawayfromhim.
An investment in knowledge always pays the best interest.‖
-Ben Franklin

―Ifpeopledidnotdosillythings,nothingintelligentwouldeverget
done.‖
-Ludwig Wittgenstein

―Exampleisnotthemainthingininfluencingothers.Itistheonlything.‖
-Albert Schweitzer

―Educationissimplythesoulofasocietyasitpassesfromonegenerationto another.‖
-G.K. Chesterton
 SPIRITUAL CORNER your self-driving car could learn how to drive safely by
watching you drive around your neighborhood.

Making progress on that vision, USC researchers have designed

शंन ो॑ दे॒ वीरे॒ भिष्टो॑ये॒ आप ो॑ िवन्तुपीे॒तय।ो॑ a system that lets robots autonomously learn complicated tasks
from a very small number of demonstrations -- even imperfect
शंय रे॒ भिस्रो॑वन्तुनः ।।यजु. ३६।१२।। ones. The paper, titled Learning from Demonstrations Using
Signal Temporal Logic, was presented at the Conference on
Robot Learning (CoRL), Nov. 18.
शब्दार्थ:
दे व ीः - The researchers' system works by evaluating the quality of
आपीः ददव्यगुणवाले दृश्यमान स्नानयोग्य तथा पानयोग्य शर र के each demonstration, so it learns from the mistakes it sees, as
well as the successes. While current state-of-art methods need
अन्दर व्यापने योग्य जल
at least 100 demonstrations to nail a specific task, this new
नीः हमार method allows robots to learn from only a handful of
अदिष्टये स्नानदिया के दलए और demonstrations. It also allows robots to learn more intuitively,
प तये पानदिया के दलए the way humans learn from each other -- you watch someone
शं िवन्तु कल्याणकार होवें। वे जल execute a task, even imperfectly, then try yourself. It doesn't
शंयोीः - have to be a "perfect" demonstration for humans to glean
knowledge from watching each other.
अदिस्रवन्तु वततमान रोगों का शमन और िाव रोगों के ियों का
पृथक्करण करें -उन्हें बाहर ि तर दोनों ओर से ररसावें-बहावें "Many machine learning and reinforcement learning systems
require large amounts of data data and hundreds of
भावार्थ:- demonstrations -- you need a human to demonstrate over and
जल का स्नान और पान करने से शर र में वततमान रोगों का शमन और over again, which is not feasible," said lead author Aniruddh
िाव रोगियों का पृथक्करण हो जाता है तथा दोनों ह प्रकार स्नान और Puranic, a Ph.D. student in computer science at the USC
पान सुखशान्तन्त को प्राप्त कराते हैं । इस प्रकार आपजनों के सङ्ग से बा Viterbi School of Engineering.
हर पापसंस्पशत का अिाव और ि तर पापताप क उपशान्तन्त होत है ।
"Also, most people don't have programming knowledge to
अध्यात्म से आपीः -
explicitly state what the robot needs to do, and a human cannot
व्यापक परमात्मा के जगत् में प्रत्यक्ष करण और अन्तरात्मा में साक्षात्
possibly demonstrate everything that a robot needs to know.
अनु िव से सच्चा सु ख और शान्तन्त क प्रान्तप्त होत है -
What if the robot encounters something it hasn't seen before?
अमृत क वर्ात होत है ।
This is a key challenge."
Learning from demonstrations
LATEST UPDATES
Learning from demonstrations is becoming increasingly
WORLD ENGINEERING popular in obtaining effective robot control policies -- which

&TECHNOLOGY
control the robot's movements -- for complex tasks. But it is
susceptible to imperfections in demonstrations and also raises
safety concerns as robots may learn unsafe or undesirable
actions.
Imagine if robots could learn from watching
Also, not all demonstrations are equal: some demonstrations
demonstrations: you could show a domestic robot how
are a better indicator of desired behavior than others and the
to do routine chores or set a dinner table. In the
quality of the demonstrations often depends on the expertise
workplace, you could train robots like new employees,
of the user providing the demonstrations.
showing them how to perform many duties. On the road,
3
 To address these issues, the researchers integrated because it reasons about sequences of true/false values for
"signal temporal logic" or STL to evaluate the quality variables, while STL allows reasoning about physical signals."
of demonstrations and automatically rank them to
Puranic, who is advised by Deshmukh, came up with the idea
create inherent rewards.
after taking a hands-on robotics class with Nikolaidis, who has
In other words, even if some parts of the been working on developing robots to learn from YouTube
demonstrations do not make any sense based on the videos. The trio decided to test it out. All three said they were
logic requirements, using this method, the robot can surprised by the extent of the system's success and the
still learn from the imperfect parts. In a way, the professors both credit Puranic for his hard work.
system is coming to its own conclusion about the
"Compared to a state-of-the-art algorithm, being used
accuracy or success of a demonstration.
extensively in many robotics applications, you see an order of
"Let's say robots learn from different types of magnitude difference in how many demonstrations are
demonstrations -- it could be a hands-on required," said Nikolaidis.
demonstration, videos, or simulations -- if I do
The system was tested using a Minecraft-style game simulator,
something that is very unsafe, standard approaches
but the researchers said the system could also learn from driving
will do one of two things: either, they will completely
simulators and eventually even videos. Next, the researchers
disregard it, or even worse, the robot will learn the
hope to try it out on real robots. They said this approach is well
wrong thing," said co-author Stefanos Nikolaidis, a
suited for applications where maps are known beforehand but
USC Viterbi assistant professor of computer science.
there are dynamic obstacles in the map: robots in household
"In contrast, in a very intelligent way, this work uses environments, warehouses or even space exploration rovers.
some common sense reasoning in the form of logic to
"If we want robots to be good teammates and help people, first
understand which parts of the demonstration are good
they need to learn and adapt to human preference very
and which parts are not. In essence, this is exactly what
efficiently," said Nikolaidis. "Our method provides that."
also humans do."
Take, for example, a driving demonstration where SRMS LATEST UPDATES
someone skips a stop sign. This would be ranked lower
by the system than a demonstration of a good driver.
But, if during this demonstration, the driver does
something intelligent -- for instance, applies their
brakes to avoid a crash -- the robot will still learn from
this smart action.
Adapting to human preferences
Signal temporal logic is an expressive mathematical
symbolic language that enables robotic reasoning about
current and future outcomes. While previous research in
this area has used "linear temporal logic," STL is
preferable in this case, said Jyo Deshmukh, a former
Toyota engineer and USC Viterbi assistant professor of
computer science .
"When we go into the world of cyber physical systems,
like robots and self-driving cars, where time is crucial,
linear temporal logic becomes a bit cumbersome,
4

Dev Uthani Ekadashi 2020:
इस ददन शाल ग्राम के साथ तुलस ज का दववाह होता है .
कादततक मास के शुक्ल पक्ष क एकादश को दे वउठन
एकादश कहा जाता है. इस बार दे वउठन एकादश बुधवार,
२६ नवंबर को है. दहंदू पंचां ग के अनुसार एक साल में कुल 24
एकादश पड़त हैं, जबदक एक माह में 2 एकादश दतदथयां
होत हैं. कादततक मास के शुक्ल पक्ष क एकादश को
दे वउठन एकादश कहते हैं. मान्यताओं के अनुसार, िगवान
दवष्णु आर्ाढ़ शुक्ल एकादश को चार माह के दलए सो जाते हैं
और कादततक शुक्ल एकादश को जागते हैं . दे वउठन
एकादश के ददन चतुमातस का अंत हो जाता है और शाद -
दववाह के काज शुरू हो जाते हैं .
दे व उत्थान ऐकादश के उपलक्ष्य में २६ नवंबर २०२० सॉय
०५:१५ बजे से राधा मंददर के प्रांगण में द पदान का आयोजन
दकया गया था। इस आयोजन में सि SRMS फैक्ल्ट , स्टाफ,
ऐवम दचदकत्सक्ल्गण आमंदित थे।
STUDENT CORNER
Smart Manufacturing 4.0 For The
‘New Normal’
The advanced technologies of Industry 4.0 automate the
collection and analysis of machine data, enabling
predictive maintenance that anticipates and corrects
potential issues and failures before they occur. Smart According to a latest study done by Market Watch, the global
Automation increases product quality and productivity cloud manufacturing market is expected to rise at a CAGR of
as manufacturers do not have to worry about downtime 19.8% over the next four years. In a cloud-based solution,
and the lost revenue that goes with it. Cloud Computing and
IoT are the enablers for smart manufacturing.
The ongoing COVID-19 pandemic forced temporary closures
on various manufacturing facilities across industries and left the
supply chain cycle around the globe severely impacted. Now
that factories have resumed operations, the key focus is on
getting back to normal production levels. It is not an easy task
but Industry 4.0’s Smart Automation innovations are coming to
the aid of companies, to not just help them return to full
production faster but to increase operational efficiency. Industry
4.0 helps transition legacy manufacturing systems to smart and
connected automated systems. The remote-enabled technology,
which uses a combination of Internet of Things (IoT), Cloud
Computing, and Real -Time Analytics, helped companies
with business continuity, as businesses stopped and then
resumed during the COVID-19 pandemic.Today we have highly
complex supply chains globally and hence it is difficult for
enterprises to monitor progress, anticipate and reduce risk.
Supply chain management applications, available through any
computer, tablet, or Smartphone, helps simplify the supply
chain by bringing together the latest in big data, cloud, graph,
and mobile technologies to deliver accurate and real-time
information.
In this industry, adaptability and flexibility in a dynamic
business environment are critical where you have to
continuously manage complex, ever-evolving supply chains to
support new customers, markets, product introductions and
variations in demand. The COVID-19 pandemic posed
challenges that were unprecedented for the manufacturing
industry. It involved learning to build abilities to effectively
respond to customer’s need and working towards developing
improved tools and processes to mitigate supply chain
disruptions.
5
productivity can be easily tracked remotely from
anywhere, while simultaneously making sure the
suppliers and customers are updated on the status as
well. Predictive models can be developed using
automated data from each production line to minimize
unforeseen downtimes and to optimize future
production schedules, thus promoting productivity and
increasing profitability. For IoT on the factory floor,
you need to have seamless connectivity and a limitless
repository of data. Cloud computing enables you to
extract, analyze, and store information that may affect
the production line.
No manufacturer can suffer long downtimes;
production needs to be maintained to stay profitable.
Prudent investments in these would go a long way
towards maintaining forward momentum. The good
news is that businesses are increasingly seeing the
benefits of Smart Automation. What they need is a
partner that can address the manufacturing complexities
of this new era and also take advantage of
advancements to bring new ideas to life.
Harshita Maurya
CS2 (2018)
Quantum Mechanics and Its Impact on
Living Processes
The use of quantum mechanics in computing has led to
significant progress in the creation of larger and more
powerful machines that can solve complex problems
that are outside the realm of capability for today’s
classical computers. IBM, Google and Intel are among
the tech conglomerates making giant strides in the race
to create truly functional quantum computers.
Presently, IBM has a 50-qubit computer, accessible
foruse over a cloud platform which falls behind
Google’s 72-qubit chip. However, Google’s version is
yet to have practical applications to test it while IBM
already has researchers working with it.
While scientists are looking into the mysterious forces
of quantum mechanics that will deliver the next
majorcomputing breakthrough, quantum forces have
been helping scientists unravel the mysteries of life
itself.
In recent years, there have been some significant
breakthroughs in the exciting field of Quantum Biology, a fast
emerging discipline that examines whether quantum mechanics
plays a fundamental part in biology and the living processes of
organisms. In theirlandmark 2014 paper, a team at University
College London found evidence that quantum mechanical
forces play a crucial role in supporting energy transfer during
photosynthesis in plants, orthe fundamental process that
ultimately powers most life on earth.
In the field of quantum computing, one of the bariers hampering
the development of quantum computers is the high instability of
qubits (the quantum equivalent of classical bits used in our
present day computers). For quantum computing to work, the
physical objects such as chips, which implement qubits, need to
be in a super cooled energy state for qubits to remain stable.
Even at that low temperature, qubits retain stability for an
extremely short time; a challenge scientists are fighting to
overcome.
As it turns out, though, living things including humans, have
been relying on quantum mechanics at room temperature inside
the wet and messy world of biology fora very long time. For
instance, scientists are baffled by the manner in which enzymes
accelerate processes. At times this acceleration is more than a
trillion-fold. In recent decades, experiments have revealed a
trick known as quantum tunneling. This is the process through
which electrons vanish from one position inside a molecule and
appearinstantly in another. Evidence of the crucial role played
by these quantum mechanical forces in many biological
processes is fast emerging. This includes in respiration, vision,
in shaping oursense of smell and even in affecting DNA
mutations that are fundamental to the existence of all life.
As this emerging field develops in the coming decade, could
these mysterious and confusing quantum forces that seem to
defy all reality, help shed light on the ultimate question of what
it means to be alive?
Ayushi Gupta
CS2 (2019)
Simulation Hypothesis
The simulation hypothesis or simulation theory is the proposal
that all of reality, including the Earth and the rest of the
universe, could in fact be an artificial simulation, such as
a computer simulation.
“If we are living in a simulation, then the cosmos that we are
6
observing is just a tiny piece of the totality of physical hypothesis. “While the world we see is in some sense ‘real,’ it
existence,” Oxford philosopher Nick Bostrom said in a is not located at the fundamental level of reality.”
2003 paper that jump-started the conversation about
what has come to be known as the simulation
That sounds like a question you might hear at a midnight
screening of "The Matrix," but lately it’s become the
subject of serious academic debate. High-profile
proponents of what’s known as the “simulation
hypothesis” include SpaceX chief Elon Musk, who
recently expounded on the idea during an interview for a
popular podcast.
“If you assume any rate of improvement at all, games will
eventually be indistinguishable from reality,” Musk said
before concluding, “We’re most likely in a simulation.”
Will we ever learn whether the simulation hypothesis is
correct? Bostrum says there’s a remote chance that one
day we might encounter a telltale glitch in the simulation.
“You could certainly imagine a scenario where a window
pops up in front of you, saying, ‘You are in a simulation;
click here for more information,’” he says. “That would
be a knock-down proof.”
Taposh Chandra Sharma
CS2 (2019)
FACULTY ARENA
Three Reasons Why COVID-19
Can Cause Silent Hypoxia
Scientists are still solving the many puzzling aspects of
how the novel coronavirus attacks the lungs and other
parts of the body. One of the biggest and most life-
threatening mysteries is how the virus causes "silent
hypoxia," a condition when oxygen levels in the body are
abnormally low, which can irreparably damage vital
organs if gone undetected for too long. Now, thanks to
computer models and comparisons with real patient data,
Boston University biomedical engineers and
collaborators from the University of Vermont have begun
to crack the mystery.
Despite experiencing dangerously low levels of oxygen,
many people infected with severe cases of COVID-19
sometimes show no symptoms of shortness of breath or
difficulty breathing. Hypoxia's ability to quietly inflict
damage is why it's been coined "silent." In coronavirus
patients, it's thought that the infection first damages the
lungs, rendering parts of them incapable of functioning
properly. Those tissues lose oxygen and stop working, no
longer infusing the blood stream with oxygen, causing
silent hypoxia. But exactly how that domino effect occurs
has not been clear until now.
"We didn't know [how this] was physiologically
possible," says Bela Suki, a BU College of Engineering
professor of biomedical engineering and of materials
science and engineering and one of the authors of the
study. Some coronavirus patients have experienced what
some experts have described as levels of blood oxygen
that are "incompatible with life." Disturbingly, Suki says,
many of these patients showed little to no signs of
abnormalities when they underwent lung scans.
7
To help get to the bottom of what causes silent hypoxia, Herrmann, a research postdoctoral associate in Suki's lab
BU biomedical engineers used computer modeling to test and the lead author of the new study.
out three different scenarios that help explain how and
Normally, the lungs perform the life-sustaining duty of
why the lungs stop providing oxygen to the bloodstream.
gas exchange, providing oxygen to every cell in the body
Their research, which has been published in Nature
as we breathe in and ridding us of carbon dioxide each
Communications, reveals that silent hypoxia is likely
time we exhale. Healthy lungs keep the blood oxygenated
caused by a combination of biological mechanisms that
at a level between 95 and 100 percent -- if it dips below
may occur simultaneously in the lungs of COVID-19
92 percent, it's a cause for concern and a
patients, according to biomedical engineer Jacob
doctor might decide to intervene with supplemental
oxygen. (Early in the coronavirus pandemic,
when clinicians first started sounding the alarm about medical scans can form inside the lungs. They found, using
silent hypoxia, oximeters flew off store shelves as many computer modeling of the lungs, that this could incite silent
people, worried that they or their family members might hypoxia, but alone it is likely not enough to cause oxygen levels
have to recover from milder cases of coronavirus at to drop as low as the levels seen in patient data.
home, wanted to be able to monitor their blood oxygen
Last, the researchers used their computer model to find out if
levels.)
COVID-19 interferes with the normal ratio of air-to-blood flow
The researchers first looked at how COVID-19 impacts that the lungs need to function normally. This type of
the lungs' ability to regulate where blood is directed. mismatched air-to-blood flow ratio is something that happens in
Normally, if areas of the lung aren't gathering much many respiratory illnesses, such as with asthma patients, Suki
oxygen due to damage from infection, the blood vessels says, and it can be a possible contributor to the severe, silent
will constrict in those areas. This is actually a good thing hypoxia that has been observed in COVID-19 patients. Their
that our lungs have evolved to do, because it forces blood models suggest that for this to be a cause of silent hypoxia, the
to instead flow through lung tissue replete with oxygen, mismatch must be happening in parts of the lung that don't
which is then circulated throughout the rest of the body. appear injured or abnormal on lung scans.
But according to Herrmann, preliminary clinical data Altogether, their findings suggest that a combination of all three
have suggested that the lungs of some COVID-19 factors are likely to be responsible for the severe cases of low
patients had lost the ability of restricting blood flow to oxygen in some COVID-19 patients. By having a better
already damaged tissue, and in contrast, were potentially understanding of these underlying mechanisms, and how the
opening up those blood vessels even more -- something combinations could vary from patient to patient, clinicians can
that is hard to see or measure on a CT scan. make more informed choices about treating patients using
measures like ventilation and supplemental oxygen. A number
Using a computational lung model, Herrmann, Suki, and
of interventions are currently being studied, including a low-
their team tested that theory, revealing that for blood
tech intervention called prone positioning that flips patients over
oxygen levels to drop to the levels observed in COVID-
onto their stomachs, allowing for the back part of the lungs to
19 patients, blood flow would indeed have to be much
pull in more oxygen and evening out the mismatched air-to-
higher than normal in areas of the lungs that can no
blood ratio.
longer gather oxygen -- contributing to low levels of
oxygen throughout the entire body, they say. "Different people respond to this virus so differently," says
Suki. For clinicians, he says it's critical to understand all the
Next, they looked at how blood clotting may impact
possible reasons why a patient's blood oxygen might be low, so
blood flow in different regions of the lung. When the
that they can decide on the proper form of treatment, including
lining of blood vessels get inflamed from COVID-19
8
infection, tiny blood clots too small to be seen on
medications that could help constrict blood vessels, bust skipping over experiments that would give redundant
blood clots, or correct a mismatched air-to-blood flow information. Helping scientists reach their goals faster with
ratio. fewer experiments also enables labs to use their limited
resources more efficiently. But how is CAMEO able to do this?
Ms. Shruti Agarwal
The Method Behind the Machine
Assistant Professor
Machine learning is a process in which computer programs can
access data and process it themselves, automatically improving
AI System Discovers Useful on their own instead of relying on repeated training. This is the
New Material basis for CAMEO, a self-learning AI that uses prediction and
uncertainty to determine which experiment to try next.
When the words "artificial intelligence" (AI) come to
mind, your first thoughts may be of super-smart As implied by its name, CAMEO looks for a useful new
computers, or robots that perform tasks without needing material by operating in a closed loop: It determines which
any help from humans. Now, a multi-institutional team experiment to run on a material, does the experiment, and
including researchers from the National Institute of collects the data. It can also ask for more information, such as
Standards and Technology (NIST) has accomplished the crystal structure of the desired material, from the scientist
something not too far off: They developed an AI before running the next experiment, which is informed by all
algorithm called CAMEO that discovered a potentially past experiments performed in the loop.
useful new material without requiring additional
training from scientists. The AI system could help reduce "The key to our experiment was that we were able to unleash
the amount of trial-and-error time scientists spend in the CAMEO on a combinatorial library where we had made a large
lab, while maximizing productivity and efficiency in their array of materials with all different compositions," said Ichiro
research. Takeuchi, a materials science and engineering researcher and
professor at the University of Maryland. In a usual
combinatorial study, every material in the array would be
The research team published their work on CAMEO
measured sequentially to look for the compound with the best
in Nature Communications.
properties. Even with a fast measurement setup, that takes a long
In the field of materials science, scientists seek to time. With CAMEO, it took only a small fraction of the usual
discover new materials that can be used in specific number of measurements to home in on the best material.
applications, such as a "metal that's light but also strong
The AI is also designed to contain knowledge of key principles,
for building a car, or one that can withstand high stresses
including knowledge of past simulations and lab experiments,
and temperatures for a jet engine," said NIST researcher
how the equipment works, and physical concepts. For example,
Aaron Gilad Kusne.
the researchers armed CAMEO with the knowledge of phase
But finding such new materials usually takes a large mapping, which describes how the arrangement of atoms in a
number of coordinated experiments and time-consuming material changes with chemical composition and temperature.
theoretical searches. If a researcher is interested in how
Understanding how atoms are arranged in a material is
a material's properties vary with different temperatures,
important in determining its properties such as how hard or how
then the researcher may need to run 10 experiments at 10
electrically insulating it is, and how well it is suited for a specific
different temperatures. But temperature is just one
application.
parameter. If there are five parameters, each with 10
values, then that researcher must run the experiment 10 "The AI is unsupervised. Many types of AI need to be trained
x 10 x 10 x 10 x 10 times, a total of 100,000 experiments. or supervised. Instead of asking it to learn physical laws, we
It's nearly impossible for a researcher to run that many encode them into the AI. You don't need a human to train the
experiments due to the years or decades it may take, AI," said Kusne.
Kusne said.
One of the best ways to figure out the structure of a material is
That's where CAMEO comes in. Short for Closed-Loop by bombarding it with X-rays, in a technique called X-ray
Autonomous System for Materials Exploration and diffraction. By identifying the angles at which the X-rays
Optimization, CAMEO can ensure that each experiment
bounce off, scientists can determine how atoms are arranged in
maximizes the scientist's knowledge and understanding, 9
a material, enabling them to figure out its crystal structure.
However, a single in-house X-ray diffraction experiment variations possible in the Ge-Sb-Te alloy system, the new
can take an hour or more. At a synchrotron facility where material GST467 discovered by CAMEO is optimal for phase-
a large machine the size of a football field accelerates change applications.
electrically charged particles at close to the speed of
Researchers wanted CAMEO to find the best Ge-Sb-Te alloy,
light, this process can take 10 seconds because the fast-
one that had the largest difference in "optical contrast" between
moving particles emit large numbers of X-rays. This is
the crystalline and amorphous states. On a DVD or Blu-ray disc,
the method used in the experiments, which were
for example, optical contrast allows a scanning laser to read the
performed at the Stanford Synchrotron Radiation Light
disc by distinguishing between regions that have high or low
source (SSRL).
reflectivity. They found that GST467 has twice the optical
The algorithm is installed on a computer that connects to contrast of ?Ge?_2 ?Sb?_2 ?Te?_5, a well-known material that's
the X-ray diffraction equipment over a data network. commonly used for DVDs. The larger contrast enables the new
CAMEO decides which material composition to study material to outperform the old material by a significant margin.
next by choosing which material the X-rays focus on to
GST467 also has applications for photonic switching devices,
investigate its atomic structure. With each new iteration,
which control the direction of light in a circuit. They can also be
CAMEO learns from past measurements and identifies
applied in neuromorphic computing, a field of study focused on
the next material to study. This allows the AI to explore
developing devices that emulate the structure and function of
how a material's composition affects its structure and
neurons in the brain, opening possibilities for new kinds of
identify the best material for the task.
computers as well as other applications such as extracting useful
"Think of this process as trying to make the perfect data from complex images.
cake," Kusne said. "You're mixing different types of
CAMEO's Wider Applications
ingredients, flour, eggs, or butter, using a variety of
recipes to make the best cake." With the AI, it's searching The researchers believe CAMEO can be used for many other
through the "recipes" or experiments to determine the materials applications. The code for CAMEO is open source and
best composition for the material. will be freely available for use by scientists and researchers.
And unlike similar machine-learning approaches, CAMEO
That approach is how CAMEO discovered the material?
discovered a useful new compound by focusing on the
Ge?_4 ?Sb?_6 ?Te?_(7,) which the group shortened to
composition-structure-property relationship of crystalline
GST467. CAMEO was given 177 potential materials to
materials. In this way, the algorithm navigated the course of
investigate, covering a large range of compositional
discovery by tracking the structural origins of a material's
recipes. To arrive at this material, CAMEO performed
functions.
19 different experimental cycles, which took 10 hours,
compared with the estimated 90 hours it would have One benefit of CAMEO is minimizing costs, since proposing,
taken a scientist with the full set of 177 materials. planning and running experiments at synchrotron facilities
requires time and money. Researchers estimate a tenfold
The New Material
reduction in time for experiments using CAMEO, since the
The material is composed of three different elements number of experiments performed can be cut by one tenth.
(germanium, antimony and tellurium, Ge-Sb-Te) and is Because the AI is running the measurements, collecting data and
a phase-change memory material, that is, it changes its performing the analysis, this also reduces the amount of
atomic structure from crystalline (solid material with knowledge a researcher needs to run the experiment. All the
atoms in designated, regular positions) to amorphous researcher must focus on is running the AI.
(solid material with atoms in random positions) when
Another benefit is providing the ability for scientists to work
quickly melted by applying heat. This type of material is
remotely. "This opens up a wave of scientists to still work and
used in electronic memory applications such as data
be productive
10 without actually being in the lab," said Apurva
storage. Although there are infinite composition
Mehta, a researcher at the SLAC National Accelerator governments and companies to implement in concrete ways.
Laboratory. This could mean that if scientists wanted to
work on research involving contagious diseases or “Moving beyond principles means understanding trade-offs and
viruses, such as COVID-19, they could do so safely and identifying the technical tools and the policy levers to address
remotely while relying on the AI to conduct the them. A need for this to be a global collaboration and engage
experiments in the lab. scientists, technologists, policymakers, and business
leaders.“This is a challenging and complex process for which
For now, researchers will continue to improve the AI and
we need all hands-on deck.”
try to make the algorithms capable of solving ever more
complex problems. "CAMEO has the intelligence of a The AI Policy Forum is designed as a yearlong process.
robot scientist, and it's built to design, run and learn from Activities associated with this effort will be distinguished by
experiments in a very efficient way," said Kusne. their focus on tangible outcomes — their engagement with key
The SSRL where the experiments took place is part of government officials at the local, national, and international
the SLAC National Accelerator Laboratory, operated by level charged with designing those public policies, and their
Stanford University for the U.S. Department of Energy deep technical grounding in the latest advances in the science of
Office of Science. SLAC researchers helped oversee the AI. The measure of success will be whether these efforts have
experiments run by CAMEO. bridged the gap between these communities, translated
principled agreement into actionable outcomes, and helped
Researchers at the University of Maryland provided the
create the conditions for deeper trust between humans and
materials used in the experiments, and researchers at the
machines.
University of Washington demonstrated the new
material in a phase-change memory device. The global collaboration will begin in late 2020 and early 2021
Ms. Neha Sharma with a series of AI Policy Forum Task Forces, chaired by MIT
Assistant Professor researchers and bringing together the world’s leading technical
and policy experts on some of the most pressing issues of AI
policy, starting with AI in finance and mobility. Further task
A Global Collaboration To Move Artificial forces throughout 2021 will convene more communities of
Intelligence Principles To Practice practice with the shared aim of designing the next chapter of AI:
Today, artificial intelligence — and the computing one that both delivers on AI’s innovative potential and responds
systems that underlie it — are more than just matters of to society’s needs.
technology; they are matters of state and society, of
governance and the public interest. The choices that Each task force will produce results that inform concrete public
technologists, policymakers, and communities make in policies and frameworks for the next chapter of AI, and help
the next few years will shape the relationship between define the roles that the academic and business communities,
machines and humans for decades to come. civil society, and governments will need to play in making it a
reality. Research from the task forces will feed into the
The rapidly increasing applicability of AI has prompted development of the AI Policy Framework, a dynamic
a number of organizations to develop high-level assessment tool that will help governments gauge their own
principles on social and ethical issues such as privacy, progress on AI policy-making goals and guide application of
fairness, bias, transparency, and accountability. Building best practices appropriate to their own national priorities.
on those broader principles, the AI Policy Forum, a
global effort convened by the MIT Stephen A. On May 6–7, 2021, MIT will host — most likely online — the
Schwarzman College of Computing, will provide an first AI Policy Forum Summit, a two-day collaborative
overarching policy framework and tools for gathering 11 to discuss the progress of the task forces towards
equipping high-level decision-makers with a deeper
understanding of the tools at their disposal — and trade-
offs to be made — to produce better public policy around
AI, and better AI systems with concern for public policy.
Then, in fall 2021, a follow-on event at MIT will bring
together leaders from across sectors and countries and,
built atop the leading research from the task forces, the
forum will provide a focal point for work to move from
AI principles to AI practice, and serve as a springboard
to global efforts to design the future of AI.
Ms. Shivangi Agarwal
Assistant Professor

12