BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
DTSTART:20190310T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:EDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20191103T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:EST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:35b39118-781d-47a7-bab1-6a2c6ffe1cb2
DTSTART;TZID=America/New_York:20190523T190000
DTEND;TZID=America/New_York:20190523T210000
CREATED:20190511T163425
DESCRIPTION:IEEE Computer Society and GBC/ACM\n7:00 PM\, Thursday\, 23 May
2019\nMIT Room 32-D463 (Star)\nThis talk will be webcast on the MIT CSAIL
Youtube channel http://www.youtube.com/channel/UCU85GtVgZzqVH64L0ThW8IA/li
ve beginning at 7pm.\n\nUnderstanding the Riemann Hypothesis\nSundar Sunda
ramurthy\n\nThere is a close connection between the roots of Riemann's zet
a function and prime numbers. Prime numbers play an important role in cryp
tography. Riemann Hypothesis is the most important unsolved problem in al
l of mathematics. It was hypothesised more than 150 years ago. The reason
it is not proven yet is because of the complex nature of analysis in the
fourth dimension. In this talk we will try to understand the hypothesis an
d give possible directions for solving it. For clarity\, we will be using
a graphical method using Labview software. We will also make a geometric c
onstruction to get the concept clear. A basic understanding of complex var
iable theory is required for the audience. Currently\, super-computer res
ources are used to prove or disprove the Riemann hypothesis using Brute fo
rce technique. The Clay Institute in Cambridge\, Massachussets has offered
1M$ in money to whoever gives a proof to this hypothesis.\n\nSundar M. Su
ndaramurthy is the Chief Technical Officer at Navin Enterprises LLC\, an e
lectronics consulting company in Massachusetts. Sundar has been a technica
l staff at MIT Lincoln Laboratory working in DSP algorithms and architectu
res. He has published many technical papers and has taught at many of the
universities in New England. He obtained his B.E. (Electronics and Communi
cations Engineering) from University of Madras in 1973\, M.S.(by research)
from Indian Institute of Technology\, Chennai in 1975 and Ph.D. from Conc
ordia University\, Montreal in 1979.\n\nThis joint meeting of the Boston C
hapter of the IEEE Computer Society and GBC/ACM will be held in MIT Room 3
2-D463 (the Star conference room on the 4th floor of the Stata Center\, bu
ilding 32 on MIT maps) . You can see it on this map of the MIT campus.\n\
nUp-to-date information about this and other talks is available online at
http://ewh.ieee.org/r1/boston/computer/. You can sign up to receive update
d status information about this talk and informational emails about future
talks at http://mailman.mit.edu/mailman/listinfo/ieee-cs\, our self-admin
istered mailing list.
LAST-MODIFIED:20190520T141010
SUMMARY:Understanding the Riemann Hypothesis
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:b79128e6-3d85-4edb-835c-fc1b146efb1c
DTSTART;TZID=America/New_York:20190528T100000
DTEND;TZID=America/New_York:20190528T110000
CREATED:20190522T132752
DESCRIPTION:Abstract: This is an exciting time for quantum computing. Over
the next few years\, intermediate-scale quantum computers with ~50-100 qub
its are expected to become practical. These computers are too small to do
error correction and they may not even capture the full power of a univers
al quantum computer. A major theoretical challenge is to understand the ca
pabilities and limitations of these devices. In order to approach this cha
llenge\, quantum supremacy experiments have been proposed as a near-term m
ilestone. The objective of quantum supremacy is to find computational prob
lems that are feasible on a small-scale quantum computer but are hard to s
imulate classically. Even though a quantum supremacy experiment may not ha
ve practical applications\, achieving it will demonstrate that quantum com
puters can outperform classical ones on\, at least\, artificially designed
computational problems. Among other proposals\, over the recent years\, t
wo sampling based quantum supremacy experiments have been proposed:\n\n(1)
The first one is based on sampling from the output of a random circuit ap
plied to a square grid of qubits. The Google quantum AI group is planning
to implement this task on a processor composed of a few (~50-100) supercon
ducting qubits. In order to argue that this sampling task is hard\, buildi
ng on previous works of Aaronson\, Arkhipov\, and others\, they conjecture
d that the output distribution of a low-depth (scaling asymptotically as t
he square root of the number of qubits) circuit is anti-concentrated meani
ng that it has nearly maximal entropy. This is known as the “anti-concen
tration” conjecture.\n\n(2) The second proposal\, known as Boson Samplin
g (Aaronson Arkhipov ’10)\, is based on linear optical experiments. A ba
seline conjecture of Boson Sampling is that it is #P-hard to approximate t
he permanent of a Gaussian matrix with zero mean and unit variance with hi
gh probability. This is known as the permanent of Gaussians conjecture.\n\
nThe first part of this thesis proves the anti-concentration conjecture fo
r random quantum circuits. The proof is an immediate consequence of our ma
in result which settles a conjecture of Brandão-Harrow-Horodecki ’12 th
at short-depth random circuits are pseudo-random. These pseudo-random quan
tum processes have many applications in algorithms\, communication\, crypt
ography as well as theoretical physics e.g. the so-called black hole infor
mation problem. This result is based on joint work with Aram Harrow (QIP 2
018).\n\nThe second part of this thesis makes progress towards the permane
nt of Gaussians conjecture and shows that the permanent of Gaussian matric
es can indeed be approximated in quasi-polynomial time with high probabili
ty if instead of zero mean one considers a nonzero but vanishing mean (~1/
polyloglog in the size of the matrix). This result finds\, to the best of
our knowledge\, the first example of a natural counting problem that is #
P-hard to compute exactly on average and #P-hard to approximate in the wor
st case but becomes easy only when approximation and average case are comb
ined. This result is based on joint work with Lior Eldar (FOCS 2018).\nThe
sis Committee: Scott Aaronson\, Ryan Williams\, Aram Harrow.\n\n
LAST-MODIFIED:20190522T134800
LOCATION:32-G575
SUMMARY:The Complexity of Sampling from a Weak Quantum Computer
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:a76f26c5-e623-45f8-a1e9-6ae1c5acead5
DTSTART;TZID=America/New_York:20190528T130000
DTEND;TZID=America/New_York:20190528T180000
CREATED:20190520T125543
DESCRIPTION:TEDxMIT events will feature talks about important and impactful
ideas by members of the broader MIT community.\n \nJoin us for the inaugu
ral TEDxMIT event at CSAIL between 1pm - 6pm on Tuesday\, May 28\, 2019 in
the Stata Center\, 32 Vassar Street\, Cambridge\, MA. This event is organ
ized by Daniela Rus and John Werner\, in collaboration with a team of unde
rgraduate students led by Stephanie Fu and Rucha Keklar. \n \nPlease regis
ter at https://tedxmit.org/ Registration is limited and will be allocate
d on a first come\, first served basis. \n \nThe Inaugural TEDxMIT event
will include talks by amazing women technologists from MIT: Barbara Liskov
\, Dava Neuman\, Hamsa Balakrishnan\, Judy Brewer\, Julie Shah\, Krystyn V
an Vliet\, Nergis Mavalvala\, Ronitt Rubinfeld\, Vivienne Sze\, and Hane L
ee. At the end of the event there will be a reception. \n\nAgenda and more
details are available at https://tedxmit.org/
LAST-MODIFIED:20190520T125543
SUMMARY:TEDxMIT
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:367e2c87-1dc2-40c6-a6c9-8469fd5600d5
DTSTART;TZID=America/New_York:20190529T160000
DTEND;TZID=America/New_York:20190529T170000
CREATED:20190520T122820
DESCRIPTION:Abstract: The resurgence of neural networks has revolutionized
artificial intelligence since 2010. Luckily for mathematicians and statist
ical physicists\, the study of large random network scaling limits\, which
can be thought of as *nonlinear* random matrix theory\, is both practical
ly important and mathematically interesting. We describe several problems
in this setting and develop a new comprehensive framework\, called “tens
or\nprograms\,” for solving these problems. This framework can be though
t of as an automatic tool to derive the behavior of computation graphs wit
h large matrices\, as used in neural network computation. It is very gener
al\, and from it we also obtain new proofs of the semicircle and the March
enko-Pastur laws. Thus\, “tensor programs” is broadly useful to linear
\nand nonlinear random matrix theory alike\, and we hope it will be adopte
d as a standard tool.\nThis talk presents the work arXiv:1902.04760.\n\n\n
LAST-MODIFIED:20190520T123123
LOCATION:32-G575
SUMMARY:Tensor Programs: A Swiss-Army Knife for Nonlinear Random Matrix The
ory of Deep Learning and Beyond
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:2242254e-1ee9-4b50-abaf-5dc559d0a5df
DTSTART;TZID=America/New_York:20190531T103000
DTEND;TZID=America/New_York:20190531T120000
CREATED:20190204T145927
DESCRIPTION:\n\nAbstract\n\nWe compute hundreds of Bitcoin private keys and
dozens of Ethereum\, Ripple\, SSH\, and HTTPS private keys by carrying ou
t cryptanalytic attacks against digital signatures contained in public blo
ckchains and Internet-wide scans. The ECDSA signature algorithm requires t
he generation of a per-message secret nonce. This nonce must be generated
perfectly uniformly\, or else an attacker can exploit the nonce biases to
compute the long-term signing key. We use a lattice-based algorithm for so
lving the hidden number problem to efficiently compute private ECDSA keys
that were used with biased signature nonces due to multiple apparent imple
mentation vulnerabilities.\n
LAST-MODIFIED:20190423T143815
LOCATION:Hewlett\, G882
SUMMARY:Nadia Heninger: Lattice Attacks against Weak ECDSA Signatures in Cr
yptocurrencies
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:b439ec3a-990b-4b56-b65b-8a521d8a8109
DTSTART;TZID=America/New_York:20190603T150000
DTEND;TZID=America/New_York:20190603T160000
CREATED:20190520T153947
DESCRIPTION:\n\nAbstract:\n\nThe advances in technology and artificial inte
lligence lead to more sophisticated system functionalities and automation.
The more complex these functionalities may become\, the more emphasis has
to be placed on intelligible systems and understandable user interactions
. In this talk\, we share the vision of “Shaper Autonomy” that enables
a new active role for humans and turns robot cars into personalized robot
cars. We also discuss how explanations support AI-driven systems\, e.g.\,
for transparency of machine learning model predictions.\n\nPresenters:\n\
nZachary Butko is an AI Project Manager at Mercedes-Benz Research & Develo
pment North America\, Inc\, Sunnyvale\, CA\, USA.\n\nKai Jardner is a Futu
rist at Daimler AG\, Sindelfingen\, Germany.\n
LAST-MODIFIED:20190520T153947
SUMMARY:Mercedes-Benz Tech Talk: Visions and Methods for Human-Machine-Inte
raction
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:c45a0c97-30e4-4567-a3f1-d04cecbf5d88
DTSTART;TZID=America/New_York:20190607T133000
DTEND;TZID=America/New_York:20190607T150000
CREATED:20190416T155834
DESCRIPTION:Congratulations CSAIL graduates! Please join us for some cake a
nd light refreshments following the commencement exercises on Friday\, Jun
e 7\, 2019 in the R&D Commons.
LAST-MODIFIED:20190416T155908
LOCATION:R&D Commons
SUMMARY:CSAIL Commencement Reception
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20190522T211455Z
UID:4a703261-44f9-4116-8e0b-889cdb14e9f2
DTSTART;TZID=America/New_York:20190620T190000
DTEND;TZID=America/New_York:20190620T210000
CREATED:20190516T175132
DESCRIPTION:IEEE Computer Society and GBC/ACM\n7:00 PM\, Thursday\, 20 June
2019\nMIT Room 32-G449 (Kiva)\n\nThis talk will be webcast on the MIT CSA
IL Youtube channel\nhttp://www.youtube.com/channel/UCYs2iUgksAhgoidZwEAimm
g/live beginning at\n7pm.\n\nWeakly Supervised Machine Learning at Industr
ial Scale\nStephen Bach\, Brown \n\nLabeling training data is one of the m
ost costly bottlenecks in developing machine learning-based applications.
Weak supervision\, using less expensive but noisier sources of supervision
than hand-labeled data\, has the potential to relax this bottleneck but i
ntroduces new challenges around managing these sources. In this talk\, I'l
l describe a new system\, Snorkel DryBell\, in production at Google for we
akly supervised machine learning at industrial scale. Snorkel DryBell buil
ds on the Snorkel framework ( snorkel.stanford.edu)\, extending it in thre
e critical aspects: flexible\, template-based ingestion of diverse organiz
ational knowledge\, cross-feature production serving\, and scalable\, samp
ling-free execution. On three classification tasks at Google\, we find tha
t Snorkel DryBell creates classifiers of comparable quality to ones traine
d with tens of thousands of hand-labeled examples\, converts non-servable
organizational resources to servable models for an average 52% performance
improvement\, and scales to millions of training examples.\n\nStephen Bac
h is an assistant professor in the computer science department at Brown Un
iversity. Previously\, he was a visiting scholar at Google\, and a postdoc
toral scholar in the computer science department at Stanford University ad
vised by Christopher Re. He received his Ph.D. in computer science from th
e University of Maryland\, where he was advised by Lise Getoor. His resear
ch focuses on statistical machine learning methods that exploit high-level
knowledge like rules and programs. Stephen's thesis on probabilistic soft
logic was recognized with the University of Maryland's Larry S. Davis Doc
toral Dissertation Award. His work on the Snorkel project for weakly super
vised machine learning was recognized with a Best of VLDB 2018 selection.\
n\nThis joint meeting of the Boston Chapter of the IEEE Computer Society a
nd GBC/ACM will be held in MIT Room 32-G449 (the Kiva conference room on t
he 4th floor of the Stata Center\, building 32 on MIT maps). You can see
it on this map of the MIT campus: \n\nUp-to-date information about this
and other talks is available online at http://ewh.ieee.org/r1/boston/compu
ter/. You can sign up to receive updated status information about this tal
k and informational emails about future talks at http://mailman.mit.edu/ma
ilman/listinfo/ieee-cs\, our self-administered mailing list.
LAST-MODIFIED:20190519T080928
SUMMARY:Weakly Supervised Machine Learning at Industrial Scale
END:VEVENT
END:VCALENDAR