BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Machine Learning @ CUED
SUMMARY:A tale of P-matrices and TripleSpinners - the unre
asonable effectiveness of structured models in non
linear embeddings - Krzysztof Choromanski\, Google
NY
DTSTART;TZID=Europe/London:20161124T110000
DTEND;TZID=Europe/London:20161124T120000
UID:TALK67544AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/67544
DESCRIPTION:In this talk we will present new generic paradigms
for conducting machine learning computations with
structured linear projections. These paradigms ca
n be applied to speed up several machine learning
algorithms based on random linear projections such
as: cross-polytope LSH techniques\, kernel approx
imations via random feature maps\, quantization me
thods using random projection trees\, several vari
ants of Johnson-Lindenstrauss transforms\, and man
y more.\n \nTheir adaptive versions can be applied
in neural network architectures to construct much
more compact and faster yet still good quality ne
ural network models. As a byproduct\, we give the
first theoretical guarantees regarding the fastest
known cross-polytope LSH methods based on the Wal
sh-Hadamard Transform.\n\nThe proposed structured
families of P-model and TripleSpin matrices cover
as special cases all structured matrices used so f
ar in this context\, but also include new structur
ed constructions not considered previously.\n\n\nB
io: \nKrzysztof Choromanski is a member of the Goo
gle Brain Robotics Team in New York. He works on s
everal problems regarding robotics and machine lea
rning such as: reinforcement learning\, control th
eory\, state estimation\, predictive state represe
ntation and Hilbert embeddings of dynamical system
s. His research interests include also\nneural net
works and the theory of structured nonlinear embed
dings. The latter are used in several machine lear
ning applications such as: compact neural networks
with fast inference\, kernel approximation techni
ques via random feature maps\, and the fastest kno
wn cross-polytope LSH algorithms. Krzysztof plays
piano and is an avid salsa dancer.
LOCATION:CBL Room BE-438
CONTACT:Adrian Weller
END:VEVENT
END:VCALENDAR