BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Tensor network from quantum simulations to machine learning
DTSTART;VALUE=DATE-TIME:20180614T010000Z
DTEND;VALUE=DATE-TIME:20180614T020000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030265@indico.cern.ch
DESCRIPTION:Speakers: Jing Chen (Flatiron Institute)\nTensor network is bo
th a theoretical and numerical tool\, which has achieved great success in
many body physics from calculating he thermodynamic property and quantum p
hase transition to simulations of black holes. As a general form of high d
imensional data structure\, tensors have been adopted in diverse branches
of data analysis\, such as in signal and image processing\, psychometric\,
quantum chemistry\, biometric\, quantum information\, back holes\, and br
ain science. Tensor network simulates the interactions between tensors and
becomes a developing powerful in these new fields. During recent years\,
tensor network numerical methods such as matrix product state (MPS) and pr
ojected entangled pair state (PEPS) has also finds its way to machine lear
ning. Besides\, the physical concept of entanglement offers a new theoret
ical approach to the design of different neural networks. \nFor example\,
we find that graphic models\, such as restricted Boltzmann machine (RBM) i
s equivalent to a specific tensor network and we can study the expression
power of the RBM.\n\nPhys. Rev. B 97\, 085104 (2018)\narXiv: 1712.04144\n\
nhttps://indico.cern.ch/event/704438/contributions/3030265/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030265/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Renormalization and hierarchical knowledge representations
DTSTART;VALUE=DATE-TIME:20180612T010000Z
DTEND;VALUE=DATE-TIME:20180612T020000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030256@indico.cern.ch
DESCRIPTION:Speakers: Cedric Beny (Hanyang University)\nOur understanding
of any given complex physical system is made of not just one\, but many th
eories which capture different aspects of the system. These theories are o
ften stitched together only in informal ways. An exception is given by ren
ormalization group techniques\, which provide formal ways of hierarchicall
y connecting descriptions at different scales.\n\nIn machine learning\, th
e various layers of a deep neural network seem to represent different leve
ls of abstraction. How does this compare to scale in renormalization? Can
one build a common information-theoretic framework underlying those techni
ques? \n\nTo approach these questions\, I compare two different renormali
zation techniques (which emerged from quantum information theory)\, and at
tempt to adapt them to unsupervised learning tasks. One approach\, MERA\,
superficially resembles a deep convolutional neural net\, while another ap
proach based on dimensional reduction yields something similar to principa
l component analysis.\n\nhttps://indico.cern.ch/event/704438/contributions
/3030256/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030256/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Quantum Computers and Machine Learning
DTSTART;VALUE=DATE-TIME:20180611T064500Z
DTEND;VALUE=DATE-TIME:20180611T073000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030255@indico.cern.ch
DESCRIPTION:Speakers: Artur Garcia Saez (Barcelona SC Center)\nI will disc
uss the two-fold relation between Quantum Computers and Machine Learning.
On one hand Quantum Computers offer new algorithms to perform training tas
ks on classical or Quantum data. On the other hand\, Machine Learning offe
rs new tools to study Quantum Matter\, and to control Quantum experiments.
\n\nhttps://indico.cern.ch/event/704438/contributions/3030255/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030255/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Machine Learning Role on High Energy Physics: A theoretical vi
ew
DTSTART;VALUE=DATE-TIME:20180615T031500Z
DTEND;VALUE=DATE-TIME:20180615T040000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030272@indico.cern.ch
DESCRIPTION:Speakers: Javier Andres Orduz Ducuara (UNAM)\nIn this talk\, I
show some concepts in computing\, Physics and Mathematics \nfocusing on H
igh Energy Physics. I share some programming languages and \ntools impleme
nted for computing the amplitudes\, decays and cross sections. \nIn partic
ular\, I explore the Two-Higgs Doublet Model and Extended Gauge Group \nMo
del and some results using Artificial Inteligence.\n\nhttps://indico.cern.
ch/event/704438/contributions/3030272/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030272/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Detection of phase transition via convolutional neural networks
DTSTART;VALUE=DATE-TIME:20180615T023000Z
DTEND;VALUE=DATE-TIME:20180615T031500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030271@indico.cern.ch
DESCRIPTION:Speakers: Akio Tomiya (Central China Normal University)\nA Con
volutional Neural Network (CNN) is designed to study correlation between t
he temperature and the spin configuration of the 2 dimensional Ising model
. \nOur CNN is able to find the characteristic feature of the phase trans
ition without prior knowledge. Also a novel order parameter on the basis
of the CNN is introduced to identify the location of the critical temperat
ure\; the result is found to be consistent with the exact value. This talk
is based on following paper\,\nJournal of the Physical Society of Japan 8
6 (6)\, 063001 (arXiv:1609.09087).\n\nhttps://indico.cern.ch/event/704438/
contributions/3030271/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030271/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Deep Learning and AdS/CFT
DTSTART;VALUE=DATE-TIME:20180615T010000Z
DTEND;VALUE=DATE-TIME:20180615T020000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030270@indico.cern.ch
DESCRIPTION:Speakers: Koji Hashimoto (Osaka University)\nWe present a deep
neural network representation of the AdS/CFT correspondence\, and demonst
rate the emergence of the bulk metric function via the learning process fo
r given data sets of response in boundary quantum field theories. The emer
gent radial direction of the bulk is identified with the depth of the laye
rs\, and the network itself is interpreted as a bulk geometry. Our network
provides a data-driven holographic modeling of strongly coupled systems.
With a scalar ϕ4 theory with unknown mass and coupling\, in unknown curve
d spacetime with a black hole horizon\, we demonstrate our deep learning (
DL) framework can determine them which fit given response data. First\, we
show that\, from boundary data generated by the AdS Schwarzschild spaceti
me\, our network can reproduce the metric. Second\, we demonstrate that ou
r network with experimental data as an input can determine the bulk metric
\, the mass and the quadratic coupling of the holographic model. As an exa
mple we use the experimental data of magnetic response of a strongly corre
lated material Sm0.6Sr0.4MnO3. This AdS/DL correspondence not only enables
gravity modeling of strongly correlated systems\, but also sheds light on
a hidden mechanism of the emerging space in both AdS and DL.\n(Work in co
llaboration with A.Tanaka\, A.Tomiya and S.Sugishita\, arXiv:1802.08313)\n
\nhttps://indico.cern.ch/event/704438/contributions/3030270/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030270/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine learning for parton distribution determination
DTSTART;VALUE=DATE-TIME:20180614T064500Z
DTEND;VALUE=DATE-TIME:20180614T073000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030269@indico.cern.ch
DESCRIPTION:Speakers: Stefano Carrazza (CERN)\nParton Distribution Functio
ns (PDFs) are a crucial ingredient for\naccurate and reliable theoretical
predictions for precision\nphenomenology at the LHC. The NNPDF approach to
the extraction of\nParton Distribution Functions relies on Monte Carlo te
chniques and\nArtificial Neural Networks to provide an unbiased determinat
ion of\nparton densities with a reliable determination of their uncertaint
ies.\nI will discuss the NNPDF methodology in general\, the latest NNPDF\n
global fit (NNPDF3.1) and then present ideas to improve the training\nmeth
odology used in the NNPDF fits and related PDFs.\n\nhttps://indico.cern.ch
/event/704438/contributions/3030269/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030269/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Toward reduction of autocorrelation in HMC by machine learning
DTSTART;VALUE=DATE-TIME:20180614T060000Z
DTEND;VALUE=DATE-TIME:20180614T064500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030268@indico.cern.ch
DESCRIPTION:Speakers: Akinori Tanaka (Riken)\nRecent development of machin
e learning (ML)\, especially deep learning is remarkable. It has been appl
ied to image recognition\, image generation and so on with very good preci
sion. From a mathematical point of view\, images are just real matrices\,
so it would be a natural idea to replace this matrices with the configurat
ions of the physical system created by numerical simulation and see what h
appens. In this talk\, I will review our attempt to reduce autocorrelation
of Hamiltonian Monte Carlo (HMC) algorithm. In addition\, I would like to
discuss a possibility of using recent sophisticated generative models lik
e VAE\, GAN to improve HMC. (work in collaboration with A. Tomiya\, arXiv:
1712.03893)\n\nhttps://indico.cern.ch/event/704438/contributions/3030268/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030268/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Reverse engineering Hamiltonian from spectrum
DTSTART;VALUE=DATE-TIME:20180614T031500Z
DTEND;VALUE=DATE-TIME:20180614T040000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030267@indico.cern.ch
DESCRIPTION:Speakers: Hiroyu Fujita (University of Tokyo)\nHandling the la
rge number of degrees of freedom with proper approximations\, namely the c
onstruction of the effective Hamiltonian is at the heart of the (condensed
matter) physics. Here we propose a simple scheme of constructing Hamilton
ians from given energy spectrum. The sparse nature of the physical Hamilto
nians allows us to formulate this as a solvable supervised learning proble
m. Taking a simple model of correlated electron systems\, we demonstrate t
he data-driven construction of its low-energy effective model. Moreover\,
we find that the same approach works for the construction of the entangle
ment Hamiltonian of a given quantum many-body state from its entanglement
spectrum. Compared to the known approach based on the full diagonalization
of the reduced density matrix\, our one is computationally much cheeper t
hus offering a way of studying the entanglement nature of large (sub)syste
ms under various boundary conditions.\n\nhttps://indico.cern.ch/event/7044
38/contributions/3030267/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030267/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Tensor Network Holography and Deep Learning
DTSTART;VALUE=DATE-TIME:20180614T023000Z
DTEND;VALUE=DATE-TIME:20180614T031500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030266@indico.cern.ch
DESCRIPTION:Speakers: Yi-Zhuang You (UCSD & Harvard)\nMotivated by the clo
se relations of the renormalization group with both the holography duality
and the deep learning\, we propose that the holographic geometry can emer
ge from deep learning the entanglement feature of a quantum many-body stat
e. We develop a concrete algorithm\, call the entanglement feature learnin
g (EFL)\, based on the random tensor network (RTN) model for the tensor ne
twork holography. We show that each RTN can be mapped to a Boltzmann machi
ne\, trained by the entanglement entropies over all subregions of a given
quantum many-body state. The goal is to construct the optimal RTN that bes
t reproduce the entanglement feature. The RTN geometry can then be interpr
eted as the emergent holographic geometry. We demonstrate the EFL algorith
m on 1D free fermion system and observe the emergence of the hyperbolic ge
ometry (AdS_3 spatial geometry) as we tune the fermion system towards the
gapless critical point (CFT_2 point).\n\nhttps://indico.cern.ch/event/7044
38/contributions/3030266/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030266/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Nelson-Seiberg theorem\, its extensions\, string realizations\
, and possible machine learning applications
DTSTART;VALUE=DATE-TIME:20180613T064500Z
DTEND;VALUE=DATE-TIME:20180613T073000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030263@indico.cern.ch
DESCRIPTION:Speakers: Zheng Sun (Sichuan University)\nThe Nelson-Seiberg t
heorem relates F-term SUSY breaking and R-symmetries in N=1 SUSY field the
ories. I will talk its several extensions including a revision to a neces
sary and sufficient condition\, discrete R-symmetries and non-Abelian R-sy
mmetries\, relation to SUSY and W=0 vacua in the string landscape\, and so
me possible machine learning applications in the searching for SUSY vacua.
\n\nhttps://indico.cern.ch/event/704438/contributions/3030263/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030263/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Faster exploration of parameter space in supersymmetry and string
theory using machine learning
DTSTART;VALUE=DATE-TIME:20180613T060000Z
DTEND;VALUE=DATE-TIME:20180613T064500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030262@indico.cern.ch
DESCRIPTION:Speakers: Sven Krippendorf (LMU Munich)\nTBA\n\nhttps://indico
.cern.ch/event/704438/contributions/3030262/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030262/
END:VEVENT
BEGIN:VEVENT
SUMMARY:On Finding Small Cosmological Constants with Deep Reinforcement Le
arning
DTSTART;VALUE=DATE-TIME:20180613T031500Z
DTEND;VALUE=DATE-TIME:20180613T040000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030261@indico.cern.ch
DESCRIPTION:Speakers: Jim Halverson (Northeastern)\nI will review the Bous
so-Polchinski model and aspects of its computational complexity. An asynch
ronous advantage actor-critic will be used to find small cosmological cons
tants.\n\nhttps://indico.cern.ch/event/704438/contributions/3030261/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030261/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Reinforcement learning in the string landscape
DTSTART;VALUE=DATE-TIME:20180613T023000Z
DTEND;VALUE=DATE-TIME:20180613T031500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030260@indico.cern.ch
DESCRIPTION:Speakers: Fabian Ruehle (Oxford)\nIn studying the string lands
cape\, we often want to find vacua with specific properties\, but do not k
now how to select the string geometry that gives rise to such vacua. For t
his reason\, we apply reinforcement learning\, a semi-supervised approach
to machine learning in which the algorithm explores the landscape autonomo
usly while being guided towards models with given properties. We illustrat
e the approach using examples from heterotic\, type II\, and F-theory.\n\n
https://indico.cern.ch/event/704438/contributions/3030260/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030260/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Algebraic geometry of the restricted Boltzmann machine
DTSTART;VALUE=DATE-TIME:20180613T010000Z
DTEND;VALUE=DATE-TIME:20180613T020000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030259@indico.cern.ch
DESCRIPTION:Speakers: Jason Morton (Pennsylvania State University)\nTBA\n\
nhttps://indico.cern.ch/event/704438/contributions/3030259/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030259/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Topological Data Analysis for Cosmology and String Theory
DTSTART;VALUE=DATE-TIME:20180612T031500Z
DTEND;VALUE=DATE-TIME:20180612T040000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030258@indico.cern.ch
DESCRIPTION:Speakers: Gary Shiu (University of Wisconsin-Madison)\nTopolog
ical data analysis (TDA) is a multi-scale approach in computational topolo
gy used to analyze the ``shape” of large datasets by identifying which h
omological characteristics persist over a range of scales. In this talk\,
I will discuss how TDA can be used to extract physics from cosmological da
tasets (e.g.\, primordial non-Gaussianities generated by cosmic inflation)
and to explore the structure of the string landscape.\n\nhttps://indico.c
ern.ch/event/704438/contributions/3030258/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030258/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Real-space renormalization group
DTSTART;VALUE=DATE-TIME:20180612T023000Z
DTEND;VALUE=DATE-TIME:20180612T031500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030257@indico.cern.ch
DESCRIPTION:Speakers: Maciej Koch-Janusz (ETH Zurich)\nPhysical systems di
ffering in their microscopic details often display strikingly similar beha
viour when probed at macroscopic scales. Those universal properties\, larg
ely determining their physical characteristics\, are revealed by the renor
malization group (RG) procedure\, which systematically retains ‘slow’
degrees of freedom and integrates out the rest. We demonstrate a machine-l
earning algorithm based on a model-independent\, information-theoretic cha
racterization of a real-space RG capable of identifying the relevant degr
ees of freedom and executing RG steps iteratively without any prior knowle
dge about the system. We apply it to classical statistical physics proble
ms in 1 and 2D: we demonstrate RG flow and extract critical exponents. We
also discuss the optimality of the procedure.\n\nhttps://indico.cern.ch/ev
ent/704438/contributions/3030257/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030257/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Neural Program Synthesis and Neural Automated Theorem Proving\, vi
a Curry-Howard Correspondence
DTSTART;VALUE=DATE-TIME:20180611T060000Z
DTEND;VALUE=DATE-TIME:20180611T064500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030243@indico.cern.ch
DESCRIPTION:Speakers: Greg Yang (Microsoft Research)\nCurry-Howard corresp
ondence is\, roughly speaking\, the observation that proving a theorem is
equivalent to writing a program. Using this principle\, I will present a u
nified survey of recent trends in the application of deep learning in prog
ram synthesis and automated theorem proving\, with commentary on their app
licability to the working mathematician and physicists.\n\nhttps://indico.
cern.ch/event/704438/contributions/3030243/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030243/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Patterns in Calabi-Yau Threefolds
DTSTART;VALUE=DATE-TIME:20180611T031500Z
DTEND;VALUE=DATE-TIME:20180611T040000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030241@indico.cern.ch
DESCRIPTION:Speakers: Vishnu Jejjala (Witwatersrand)\nTBA\n\nhttps://indic
o.cern.ch/event/704438/contributions/3030241/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030241/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A Generalized Construction of Calabi-Yau Manifolds and Mirror Symm
etry
DTSTART;VALUE=DATE-TIME:20180611T023000Z
DTEND;VALUE=DATE-TIME:20180611T031500Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030240@indico.cern.ch
DESCRIPTION:Speakers: Per Berglund (New Hampshire)\nWe extend the construc
tion of Calabi-Yau manifolds to hypersurfaces in non-Fano toric varieties.
The associated non-reflexive polytopes provide a generalization of Batyre
v’s original work\, allowing us to construct new pairs of mirror manifol
ds. In particular\, this allows us to find new K3-fibered Calabi-Yau manif
olds\, relevant for string compactifications.\n\nhttps://indico.cern.ch/ev
ent/704438/contributions/3030240/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030240/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Learning and Lie Groups
DTSTART;VALUE=DATE-TIME:20180611T010000Z
DTEND;VALUE=DATE-TIME:20180611T020000Z
DTSTAMP;VALUE=DATE-TIME:20200810T125941Z
UID:indico-contribution-704438-3030239@indico.cern.ch
DESCRIPTION:Speakers: Gregory S. Chirikjian (Johns Hopkins University)\nMa
chine learning methods are mostly based on calculus and probability and st
atistics on Euclidean spaces.\nHowever\, many interesting problems can be
articulated as learning in lower dimensional embedded manifolds\nand on Li
e groups. This talk reviews how learning and Lie groups fit together\, and
how the machine learning community can benefit from modern mathematical d
evelopments. The topics include:\n\n•Introduction to Calculus on Lie Gro
ups (Differential Operators\, Integration)\n•Probability on Lie Groups (
Convolution\, Fourier Analysis\, Diffusion Equations)\n•Application 1: W
orkspace Generation and Inverse Kinematics of Highly Articulated Robotic M
anipulators\n•Application 2: Pose Distributions for Mobile Robots\n•Ap
plication 2: Lie-Theoretic Invariances in Image Processing and Computer Vi
sion\n•Application 3: Coset-Spaces of Lie Groups by Discrete Subgroups i
n Crystallography\n•Prospects for the Future\n\nhttps://indico.cern.ch/e
vent/704438/contributions/3030239/
LOCATION:Tsinghua Sanya International Mathematics Forum
URL:https://indico.cern.ch/event/704438/contributions/3030239/
END:VEVENT
END:VCALENDAR