From ccanonne at stanford.edu Wed Nov 1 10:35:07 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 1 Nov 2017 10:35:07 -0700
Subject: [theory-seminar] Interested? TCS+ talk: Wednesday, November 8,
Ola Svensson, EPFL
In-Reply-To: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
References: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
Message-ID: <46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
Hi everyone,
I am not sure how familiar you are with TCS+ [1], a
> carbon-free dissemination of ideas across the globe
so here is the spiel:
Talks are given online, every other week. People can attend them from
their home, university, or ski resort, and interact with the speaker;
afterwards, the talks are archived online and can be watched and watched
again.
The next TCS+ talk (see details below) will take place next Wednesday,
at 10am "our time": Ola Svensson will be speaking about the recent
breakthrough he and his coauthors obtained on a O(1)-approximation to
the ATSP.
If you're interested, let me know! I can reserve a "seat" in the talk
(to allow us to ask questions), and make sure there is some breakfast
around for us to chomp intelligently while listening to the talk.
Best,
-- Cl?ment
[1] https://sites.google.com/site/plustcs/home
-------- Forwarded Message --------
Subject: [Tcsplus_announce] TCS+ talk: Wednesday, November 8, Ola
Svensson, EPFL
Date: Wed, 1 Nov 2017 07:58:30 -0700
From: Thomas vidick
To: tcsplus_announce at cs.nyu.edu
Dear TCS+ followers,
Our next talk will take place this coming Wednesday, November 8th at
1:00 PM Eastern Time (10:00 AM Pacific Time, 19:00 Central European
Time, 18:00 UTC). Ola Svensson from EPFL will speak about his recent
algorithmic breakthrough with Tarnawski and V?gh giving "A
Constant-factor Approximation Algorithm for the Asymmetric Traveling
Salesman Problem" (abstract below).
Please sign up on the online form at
https://sites.google.com/site/plustcs/livetalk/live-seat-reservation if
you wish to join the talk as a group.
Hoping to see you all there,
The organizers
-------------------------------
Speaker: Ola Svensson (EPFL)
Title: A Constant-factor Approximation Algorithm for the Asymmetric
Traveling Salesman Problem
Abstract: We give a constant-factor approximation algorithm for the
asymmetric traveling salesman problem. Our approximation guarantee is
analyzed with respect to the standard LP relaxation, and thus our result
confirms the conjectured constant integrality gap of that relaxation.
Our techniques build upon the constant-factor approximation algorithm
for the special case of node-weighted metrics. Specifically, we give a
generic reduction to structured instances that resemble but are more
general than those arising from node-weighted metrics. For those
instances, we then solve Local-Connectivity ATSP, a problem known to be
equivalent (in terms of constant-factor approximation) to the asymmetric
traveling salesman problem.
This is joint work with Jakub Tarnawski and L?szl? V?gh.
From ccanonne at stanford.edu Wed Nov 1 13:22:07 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 1 Nov 2017 13:22:07 -0700
Subject: [theory-seminar] Interested? TCS+ talk: Wednesday, November 8,
Ola Svensson, EPFL
In-Reply-To: <46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
References: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
<46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
Message-ID:
Hi,
So, I got a strictly positive number of replies about it, and reserved a
spot for the talk.
I'll send more details this weekend, but in short: next Wednesday, 10am,
there'll be a group viewing somewhere.
Best,
Cl?ment
On 11/01/2017 10:35 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> I am not sure how familiar you are with TCS+ [1], a
>> carbon-free dissemination of ideas across the globe
> so here is the spiel:
>
> Talks are given online, every other week. People can attend them from
> their home, university, or ski resort, and interact with the speaker;
> afterwards, the talks are archived online and can be watched and watched
> again.
>
> The next TCS+ talk (see details below) will take place next Wednesday,
> at 10am "our time": Ola Svensson will be speaking about the recent
> breakthrough he and his coauthors obtained on a O(1)-approximation to
> the ATSP.
>
> If you're interested, let me know! I can reserve a "seat" in the talk
> (to allow us to ask questions), and make sure there is some breakfast
> around for us to chomp intelligently while listening to the talk.
>
> Best,
>
> -- Cl?ment
>
> [1] https://sites.google.com/site/plustcs/home
>
>
> -------- Forwarded Message --------
> Subject: [Tcsplus_announce] TCS+ talk: Wednesday, November 8, Ola
> Svensson, EPFL
> Date: Wed, 1 Nov 2017 07:58:30 -0700
> From: Thomas vidick
> To: tcsplus_announce at cs.nyu.edu
>
> Dear TCS+ followers,
>
> Our next talk will take place this coming Wednesday, November 8th at
> 1:00 PM Eastern Time (10:00 AM Pacific Time, 19:00 Central European
> Time, 18:00 UTC). Ola Svensson from EPFL will speak about his recent
> algorithmic breakthrough with Tarnawski and V?gh giving "A
> Constant-factor Approximation Algorithm for the Asymmetric Traveling
> Salesman Problem" (abstract below).
>
> Please sign up on the online form at
> https://sites.google.com/site/plustcs/livetalk/live-seat-reservation if
> you wish to join the talk as a group.
>
> Hoping to see you all there,
>
> The organizers
>
> -------------------------------
> Speaker: Ola Svensson (EPFL)
> Title: A Constant-factor Approximation Algorithm for the Asymmetric
> Traveling Salesman Problem
>
> Abstract: We give a constant-factor approximation algorithm for the
> asymmetric traveling salesman problem. Our approximation guarantee is
> analyzed with respect to the standard LP relaxation, and thus our result
> confirms the conjectured constant integrality gap of that relaxation.
>
> Our techniques build upon the constant-factor approximation algorithm
> for the special case of node-weighted metrics. Specifically, we give a
> generic reduction to structured instances that resemble but are more
> general than those arising from node-weighted metrics. For those
> instances, we then solve Local-Connectivity ATSP, a problem known to be
> equivalent (in terms of constant-factor approximation) to the asymmetric
> traveling salesman problem.
>
> This is joint work with Jakub Tarnawski and L?szl? V?gh.
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From silas at stanford.edu Thu Nov 2 11:58:27 2017
From: silas at stanford.edu (Shashwat Silas)
Date: Thu, 2 Nov 2017 18:58:27 +0000
Subject: [theory-seminar] Theory Seminar Today
Message-ID:
Hi All,
There will be a theory seminar today at 415pm at Gates 463A. Looking forward to seeing you there!
Avishay Tal (Stanford)
Computing Requires Larger Formulas than Approximating
A de-Morgan formula over Boolean variables x_1, ..., x_n is a binary tree whose internal nodes are marked with AND or OR gates and whose leaves are marked with variables or their negation. We define the size of the formula as the number of leaves in it. Proving that some explicit function (in P or NP) requires large formulas is a central open question in computational complexity.
In this work, we introduce a size-amplification hardness reduction for de-Morgan formulas. We show that average-case hardness implies worst-case hardness for a larger size. More precisely, if a function f cannot be computed correctly on more than 1/2 + eps of the inputs by any formula of size s, then computing f correctly on all inputs requires size ~s*log(1/eps). The tradeoff is essentially tight. Quite surprisingly, the proof relies on a result from quantum query complexity by Reichardt [SODA, 2011].
As an application, we improve the best known formula size lower bounds for explicit functions by logarithmic factors to ~n^3/log(n). In addition, we propose candidates for explicit functions that we believe have formula size ~n^4, and prove non trivial super-quadratic formula size lower bounds for them using our reduction.
Thanks,
Shashwat
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From marykw at stanford.edu Thu Nov 2 16:31:22 2017
From: marykw at stanford.edu (Mary Wootters)
Date: Thu, 2 Nov 2017 16:31:22 -0700
Subject: [theory-seminar] Theory Seminar Today: ROOM CHANGE
In-Reply-To:
References:
Message-ID:
On Nov 2, 2017 4:16 PM, "Mary Wootters" wrote:
Hi everyone,
The theory seminar today will be in the lab, 463 instead of 463A, starting
a bit late (aka now). Sorry for the room change and see you there!
--Mary
On Thu, Nov 2, 2017 at 11:58 AM, Shashwat Silas wrote:
> Hi All,
>
>
> There will be a theory seminar today at 415pm at Gates 463A. Looking
> forward to seeing you there!
>
>
> Avishay Tal (Stanford) Computing Requires Larger Formulas than
> Approximating
>
> A de-Morgan formula over Boolean variables x_1, ..., x_n is a binary tree
> whose internal nodes are marked with AND or OR gates and whose leaves are
> marked with variables or their negation. We define the size of the formula
> as the number of leaves in it. Proving that some explicit function (in P or
> NP) requires large formulas is a central open question in computational
> complexity.
>
> In this work, we introduce a size-amplification hardness reduction for
> de-Morgan formulas. We show that average-case hardness implies worst-case
> hardness for a larger size. More precisely, if a function f cannot be
> computed correctly on more than 1/2 + eps of the inputs by any formula of
> size s, then computing f correctly on all inputs requires size
> ~s*log(1/eps). The tradeoff is essentially tight. Quite surprisingly, the
> proof relies on a result from quantum query complexity by Reichardt [SODA,
> 2011].
>
> As an application, we improve the best known formula size lower bounds for
> explicit functions by logarithmic factors to ~n^3/log(n). In addition, we
> propose candidates for explicit functions that we believe have formula size
> ~n^4, and prove non trivial super-quadratic formula size lower bounds for
> them using our reduction.
>
>
> Thanks,
>
> Shashwat
>
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From gvaliant at cs.stanford.edu Fri Nov 3 07:39:13 2017
From: gvaliant at cs.stanford.edu (Gregory Valiant)
Date: Fri, 3 Nov 2017 07:39:13 -0700
Subject: [theory-seminar] post-deadline celebration today
Message-ID:
Hi Friends,
As usual, we will be having a post-deadline celebration. The 2pm STOC
deadline is a little early this year, but come to 463 (or my office, if 463
is occupied) for some post-deadline refreshments.
Cheers,
-g
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From marykw at stanford.edu Sun Nov 5 15:56:54 2017
From: marykw at stanford.edu (Mary Wootters)
Date: Sun, 5 Nov 2017 15:56:54 -0800
Subject: [theory-seminar] OR Seminar Nov. 6th by Amir Ali Ahmadi (Princeton,
ORFE)
Message-ID:
Hi all,
The OR seminar tomorrow might be of interest to theory folks -- see below
for details.
--Mary
Title: Polynomial Optimization and Dynamical Systems
Speaker: Amir Ali Ahmadi (Princeton, ORFE)
Time: Monday, Nov. 6th, 4:30 to 5:30 PM
Location: Spilker 317
Abstract: In recent years, there has been a surge of exciting research
activity at the interface of optimization (in particular polynomial,
semidefinite, and sum of squares optimization) and the theory of dynamical
systems. In this talk, we focus on two of our current research directions
that are at this interface. In part (i), we propose more scalable
alternatives to sum of squares optimization and show how they impact
verification problems in control and robotics, as well as some classic
questions in polynomial optimization and statistics. Our new algorithms do
not rely on semidefinite programming, but instead use linear programming,
or second-order cone programming, or are altogether free of optimization.
In particular, we present the first Positivstellensatz that certifies
infeasibility of a set of polynomial inequalities simply by multiplying
certain fixed polynomials together and checking nonnegativity of the
coefficients of the resulting product.
In part (ii), we introduce a new class of optimization problems whose
constraints are imposed by trajectories of a dynamical system. As a
concrete example, we consider the problem of optimizing a linear function
over the set of initial conditions that forever remain inside a given
polyhedron under the action of a linear, or a switched linear, dynamical
system. We present a hierarchy of linear and semidefinite programs that
respectively lower and upper bound the optimal value of such problems to
arbitrary accuracy.
Speaker Bio: Amir Ali Ahmadi ( http://aaa.princeton.edu/ ) is an Assistant
Professor at the Department of Operations Research and Financial
Engineering at Princeton University and an Associated Faculty member of the
Program in Applied and Computational Mathematics, the Department of
Computer Science, and the Department of Mechanical and Aerospace
Engineering. Amir Ali received his PhD in EECS from MIT and was a Goldstine
Fellow at the IBM Watson Research Center prior to joining Princeton. His
research interests are in optimization theory, computational aspects of
dynamics and control, and algorithms and complexity. Amir Ali's
distinctions include the Sloan Fellowship in Computer Science, the NSF
CAREER Award, the AFOSR Young Investigator Award, the DARPA Faculty Award,
the Google Faculty Award, the Howard B. Wentz Junior Faculty Award of
Princeton University, the Goldstine Fellowship of IBM Research, and the
Oberwolfach Fellowship of the NSF. His undergraduate course at Princeton
(ORF 363, ``Computing and Optimization'') has received the 2017 Excellence
in Teaching of Operations Research Award of the Institute for Industrial
and Systems Engineers and the 2017 Phi Beta Kappa Award for Excellence in
Undergraduate Teaching at Princeton University. Amir Ali is also the
recipient of a number of best-paper awards, including the INFORMS Computing
Society Prize (for best series of papers at the interface of operations
research and computer science), the Best Conference Paper Award of the IEEE
International Conference on Robotics and Automation, and the prize for one
of two most outstanding papers published in the SIAM Journal on Control and
Optimization in 2013-2015.
This seminar series is supported through the generosity of Adriana
Diener-Veinott and Infanger Investment Technologies.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Mon Nov 6 10:01:24 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Mon, 6 Nov 2017 10:01:24 -0800
Subject: [theory-seminar] Interested? TCS+ talk: Wednesday, November 8,
Ola Svensson, EPFL
In-Reply-To: <46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
References: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
<46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
Message-ID:
Hi everyone,
Following my previous email, and the replies I got shoing that there is
a clearly non-zero demand for either the talk or the breakfast (we may
never know): we'll be watching the talk this coming Wednesday, at
**10am**, in the usual seminar room.
There will be edible items from Tuts Bakery & Caf?.
See you then (and there)!
Cl?ment
On 11/01/2017 10:35 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> I am not sure how familiar you are with TCS+ [1], a
>> carbon-free dissemination of ideas across the globe
> so here is the spiel:
>
> Talks are given online, every other week. People can attend them from
> their home, university, or ski resort, and interact with the speaker;
> afterwards, the talks are archived online and can be watched and watched
> again.
>
> The next TCS+ talk (see details below) will take place next Wednesday,
> at 10am "our time": Ola Svensson will be speaking about the recent
> breakthrough he and his coauthors obtained on a O(1)-approximation to
> the ATSP.
>
> If you're interested, let me know! I can reserve a "seat" in the talk
> (to allow us to ask questions), and make sure there is some breakfast
> around for us to chomp intelligently while listening to the talk.
>
> Best,
>
> -- Cl?ment
>
> [1] https://sites.google.com/site/plustcs/home
>
>
> -------- Forwarded Message --------
> Subject: [Tcsplus_announce] TCS+ talk: Wednesday, November 8, Ola
> Svensson, EPFL
> Date: Wed, 1 Nov 2017 07:58:30 -0700
> From: Thomas vidick
> To: tcsplus_announce at cs.nyu.edu
>
> Dear TCS+ followers,
>
> Our next talk will take place this coming Wednesday, November 8th at
> 1:00 PM Eastern Time (10:00 AM Pacific Time, 19:00 Central European
> Time, 18:00 UTC). Ola Svensson from EPFL will speak about his recent
> algorithmic breakthrough with Tarnawski and V?gh giving "A
> Constant-factor Approximation Algorithm for the Asymmetric Traveling
> Salesman Problem" (abstract below).
>
> Please sign up on the online form at
> https://sites.google.com/site/plustcs/livetalk/live-seat-reservation if
> you wish to join the talk as a group.
>
> Hoping to see you all there,
>
> The organizers
>
> -------------------------------
> Speaker: Ola Svensson (EPFL)
> Title: A Constant-factor Approximation Algorithm for the Asymmetric
> Traveling Salesman Problem
>
> Abstract: We give a constant-factor approximation algorithm for the
> asymmetric traveling salesman problem. Our approximation guarantee is
> analyzed with respect to the standard LP relaxation, and thus our result
> confirms the conjectured constant integrality gap of that relaxation.
>
> Our techniques build upon the constant-factor approximation algorithm
> for the special case of node-weighted metrics. Specifically, we give a
> generic reduction to structured instances that resemble but are more
> general than those arising from node-weighted metrics. For those
> instances, we then solve Local-Connectivity ATSP, a problem known to be
> equivalent (in terms of constant-factor approximation) to the asymmetric
> traveling salesman problem.
>
> This is joint work with Jakub Tarnawski and L?szl? V?gh.
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From silas at stanford.edu Mon Nov 6 13:03:49 2017
From: silas at stanford.edu (Shashwat Silas)
Date: Mon, 6 Nov 2017 21:03:49 +0000
Subject: [theory-seminar] Theory Seminar on Thursday
Message-ID:
Hi all,
Theory Seminar will be on Thursday at 415pm in Gates 463A as usual. Hope you to see you there!
Clement Canonne (Stanford)
When Fourier SIIRVs: Fourier-Based Testing for Families of Distributions
We study the general problem of testing whether an unknown discrete probability distribution belongs to a specified family of distributions. More specifically, given a distribution family PP and sample access to an unknown discrete distribution pp, we want to distinguish (with high probability) between the case that pp is in PP and the case that pp is eps-far, in total variation distance, from every distribution in PP. This is the archetypal hypothesis testing problem that has received significant attention in statistics and, over the past decade and roughly a half, in theoretical computer science.
Of course, the sample complexity of this general inference task depends on the underlying family PP. The gold standard in distribution testing is to design sample-optimal and computationally efficient algorithms for this task, as a function of PP. The main contribution of this work is a simple and general testing technique that is applicable to /all/ distribution families, and is particularly suited to those whose /Fourier spectrum/ satisfies a certain approximate /sparsity/ property. To the best of our knowledge, ours is the first use of the Fourier transform in the context of distribution testing.
We apply our Fourier-based framework to obtain near sample-optimal and computationally efficient testers for the following fundamental distribution families: Sums of Independent Integer Random Variables (SIIRVs), Poisson Multinomial Distributions (PMDs), and Discrete Log-Concave Distributions. For the first two, ours are the first non-trivial testers in the literature, vastly generalizing previous work on testing Poisson Binomial Distributions. For the third, our tester improves on prior work in both sample and time complexity.
Joint work with Ilias Diakonikolas (USC) and Alistair Stewart (USC).
Thanks,
Shashwat
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From hongyang at cs.stanford.edu Tue Nov 7 16:18:16 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Tue, 7 Nov 2017 16:18:16 -0800
Subject: [theory-seminar] [theory-lunch] Josh Wang on Hierarchical Clustering
Message-ID:
Hi Everyone
This Thursday Josh will tell us about "Approximation Bounds for
Hierarchical Clustering" (see abstract below).
?As usual, we meet from 12:00pm to 1:00pm, at Gates 463A.
===============================
Title: Approximation Bounds for Hierarchical Clustering
Abstract: Hierarchical clustering is a data analysis method that has been
used for decades. Despite its widespread use, the method has an
underdeveloped analytical foundation. Having a well understood foundation
would both support the currently used methods and help guide future
improvements. We give an analytic framework to better understand
observations seen in practice, by considering the dual of a problem
framework for hierarchical clustering introduced by Dasgupta. Our main
result is that one of the most popular algorithms used in practice, average
linkage agglomerative clustering, has a small constant approximation ratio
for this objective. Furthermore, we establish that using bisecting k-means
divisive clustering has a very poor lower bound on its approximation ratio
for the same objective. However, we show that there are divisive algorithms
that perform well with respect to this objective by giving two constant
approximation algorithms. This is some of the first work to establish
guarantees on widely used hierarchical algorithms for a natural objective
function. This objective and analysis give insight into what these popular
algorithms are optimizing and when they will perform well.
===============================
B
?est,
Hongyang
?
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From hongyang at cs.stanford.edu Tue Nov 7 16:18:16 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Tue, 7 Nov 2017 16:18:16 -0800
Subject: [theory-seminar] [theory-lunch] Josh Wang on Hierarchical Clustering
Message-ID:
Hi Everyone
This Thursday Josh will tell us about "Approximation Bounds for
Hierarchical Clustering" (see abstract below).
?As usual, we meet from 12:00pm to 1:00pm, at Gates 463A.
===============================
Title: Approximation Bounds for Hierarchical Clustering
Abstract: Hierarchical clustering is a data analysis method that has been
used for decades. Despite its widespread use, the method has an
underdeveloped analytical foundation. Having a well understood foundation
would both support the currently used methods and help guide future
improvements. We give an analytic framework to better understand
observations seen in practice, by considering the dual of a problem
framework for hierarchical clustering introduced by Dasgupta. Our main
result is that one of the most popular algorithms used in practice, average
linkage agglomerative clustering, has a small constant approximation ratio
for this objective. Furthermore, we establish that using bisecting k-means
divisive clustering has a very poor lower bound on its approximation ratio
for the same objective. However, we show that there are divisive algorithms
that perform well with respect to this objective by giving two constant
approximation algorithms. This is some of the first work to establish
guarantees on widely used hierarchical algorithms for a natural objective
function. This objective and analysis give insight into what these popular
algorithms are optimizing and when they will perform well.
===============================
B
?est,
Hongyang
?
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Wed Nov 8 09:24:15 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 8 Nov 2017 09:24:15 -0800
Subject: [theory-seminar] Interested? TCS+ talk: Wednesday, November 8,
Ola Svensson, EPFL
In-Reply-To:
References: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
<46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
Message-ID:
Reminder: this is in 30mn in 463A!
Cl?ment
On 11/06/2017 10:01 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> Following my previous email, and the replies I got shoing that there is
> a clearly non-zero demand for either the talk or the breakfast (we may
> never know): we'll be watching the talk this coming Wednesday, at
> **10am**, in the usual seminar room.
>
> There will be edible items from Tuts Bakery & Caf?.
>
> See you then (and there)!
>
> Cl?ment
> On 11/01/2017 10:35 AM, Cl?ment Canonne wrote:
>> Hi everyone,
>>
>> I am not sure how familiar you are with TCS+ [1], a
>>> carbon-free dissemination of ideas across the globe
>> so here is the spiel:
>>
>> Talks are given online, every other week. People can attend them from
>> their home, university, or ski resort, and interact with the speaker;
>> afterwards, the talks are archived online and can be watched and
>> watched again.
>>
>> The next TCS+ talk (see details below) will take place next Wednesday,
>> at 10am "our time": Ola Svensson will be speaking about the recent
>> breakthrough he and his coauthors obtained on a O(1)-approximation to
>> the ATSP.
>>
>> If you're interested, let me know! I can reserve a "seat" in the talk
>> (to allow us to ask questions), and make sure there is some breakfast
>> around for us to chomp intelligently while listening to the talk.
>>
>> Best,
>>
>> -- Cl?ment
>>
>> [1] https://sites.google.com/site/plustcs/home
>>
>>
>> -------- Forwarded Message --------
>> Subject: [Tcsplus_announce] TCS+ talk: Wednesday, November 8, Ola
>> Svensson, EPFL
>> Date: Wed, 1 Nov 2017 07:58:30 -0700
>> From: Thomas vidick
>> To: tcsplus_announce at cs.nyu.edu
>>
>> Dear TCS+ followers,
>>
>> Our next talk will take place this coming Wednesday, November 8th at
>> 1:00 PM Eastern Time (10:00 AM Pacific Time, 19:00 Central European
>> Time, 18:00 UTC). Ola Svensson from EPFL will speak about his recent
>> algorithmic breakthrough with Tarnawski and V?gh giving "A
>> Constant-factor Approximation Algorithm for the Asymmetric Traveling
>> Salesman Problem" (abstract below).
>>
>> Please sign up on the online form at
>> https://sites.google.com/site/plustcs/livetalk/live-seat-reservation
>> if you wish to join the talk as a group.
>>
>> Hoping to see you all there,
>>
>> The organizers
>>
>> -------------------------------
>> Speaker: Ola Svensson (EPFL)
>> Title: A Constant-factor Approximation Algorithm for the Asymmetric
>> Traveling Salesman Problem
>>
>> Abstract: We give a constant-factor approximation algorithm for the
>> asymmetric traveling salesman problem. Our approximation guarantee is
>> analyzed with respect to the standard LP relaxation, and thus our
>> result confirms the conjectured constant integrality gap of that
>> relaxation.
>>
>> Our techniques build upon the constant-factor approximation algorithm
>> for the special case of node-weighted metrics. Specifically, we give a
>> generic reduction to structured instances that resemble but are more
>> general than those arising from node-weighted metrics. For those
>> instances, we then solve Local-Connectivity ATSP, a problem known to
>> be equivalent (in terms of constant-factor approximation) to the
>> asymmetric traveling salesman problem.
>>
>> This is joint work with Jakub Tarnawski and L?szl? V?gh.
>>
>> _______________________________________________
>> theory-seminar mailing list
>> theory-seminar at lists.stanford.edu
>> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From ccanonne at stanford.edu Wed Nov 8 11:53:59 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 8 Nov 2017 11:53:59 -0800
Subject: [theory-seminar] Interested? TCS+ talk: Wednesday, November 8,
Ola Svensson, EPFL
In-Reply-To:
References: <9d363c1b-f6ed-f8e6-8a7e-bdadf8bf9424@cms.caltech.edu>
<46ed8d74-bda6-fd4c-94a4-7c703620a592@stanford.edu>
Message-ID: <87a4652c-8773-19f0-72b1-31844ad3a6b2@stanford.edu>
Hi again,
In case you missed the talk this morning, or simply want more of it already:
(1) there still *may* be croissants left in the pantry
(2) the talk is online and available for future and repeated viewing at
https://sites.google.com/site/plustcs/past-talks/20171108olasvenssonepfl
(along with the slides)
Best,
Cl?ment
On 11/06/2017 10:01 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> Following my previous email, and the replies I got shoing that there is
> a clearly non-zero demand for either the talk or the breakfast (we may
> never know): we'll be watching the talk this coming Wednesday, at
> **10am**, in the usual seminar room.
>
> There will be edible items from Tuts Bakery & Caf?.
>
> See you then (and there)!
>
> Cl?ment
> On 11/01/2017 10:35 AM, Cl?ment Canonne wrote:
>> Hi everyone,
>>
>> I am not sure how familiar you are with TCS+ [1], a
>>> carbon-free dissemination of ideas across the globe
>> so here is the spiel:
>>
>> Talks are given online, every other week. People can attend them from
>> their home, university, or ski resort, and interact with the speaker;
>> afterwards, the talks are archived online and can be watched and
>> watched again.
>>
>> The next TCS+ talk (see details below) will take place next Wednesday,
>> at 10am "our time": Ola Svensson will be speaking about the recent
>> breakthrough he and his coauthors obtained on a O(1)-approximation to
>> the ATSP.
>>
>> If you're interested, let me know! I can reserve a "seat" in the talk
>> (to allow us to ask questions), and make sure there is some breakfast
>> around for us to chomp intelligently while listening to the talk.
>>
>> Best,
>>
>> -- Cl?ment
>>
>> [1] https://sites.google.com/site/plustcs/home
>>
>>
>> -------- Forwarded Message --------
>> Subject: [Tcsplus_announce] TCS+ talk: Wednesday, November 8, Ola
>> Svensson, EPFL
>> Date: Wed, 1 Nov 2017 07:58:30 -0700
>> From: Thomas vidick
>> To: tcsplus_announce at cs.nyu.edu
>>
>> Dear TCS+ followers,
>>
>> Our next talk will take place this coming Wednesday, November 8th at
>> 1:00 PM Eastern Time (10:00 AM Pacific Time, 19:00 Central European
>> Time, 18:00 UTC). Ola Svensson from EPFL will speak about his recent
>> algorithmic breakthrough with Tarnawski and V?gh giving "A
>> Constant-factor Approximation Algorithm for the Asymmetric Traveling
>> Salesman Problem" (abstract below).
>>
>> Please sign up on the online form at
>> https://sites.google.com/site/plustcs/livetalk/live-seat-reservation
>> if you wish to join the talk as a group.
>>
>> Hoping to see you all there,
>>
>> The organizers
>>
>> -------------------------------
>> Speaker: Ola Svensson (EPFL)
>> Title: A Constant-factor Approximation Algorithm for the Asymmetric
>> Traveling Salesman Problem
>>
>> Abstract: We give a constant-factor approximation algorithm for the
>> asymmetric traveling salesman problem. Our approximation guarantee is
>> analyzed with respect to the standard LP relaxation, and thus our
>> result confirms the conjectured constant integrality gap of that
>> relaxation.
>>
>> Our techniques build upon the constant-factor approximation algorithm
>> for the special case of node-weighted metrics. Specifically, we give a
>> generic reduction to structured instances that resemble but are more
>> general than those arising from node-weighted metrics. For those
>> instances, we then solve Local-Connectivity ATSP, a problem known to
>> be equivalent (in terms of constant-factor approximation) to the
>> asymmetric traveling salesman problem.
>>
>> This is joint work with Jakub Tarnawski and L?szl? V?gh.
>>
>> _______________________________________________
>> theory-seminar mailing list
>> theory-seminar at lists.stanford.edu
>> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From moses at cs.stanford.edu Wed Nov 8 13:28:10 2017
From: moses at cs.stanford.edu (Moses Charikar)
Date: Wed, 8 Nov 2017 13:28:10 -0800
Subject: [theory-seminar] Fwd: [ML lunch] Fwd: [RAIN] UPDATE: Joint GSB/RAIN
seminar - Jon Kleinberg - today at 3:45 PM
In-Reply-To:
References:
Message-ID:
Theory folks,
Jon Kleinberg is speaking over at the GSB today at 3:45 in the
MBA Class of 1968 building, room C105. Details below.
Cheers,
Moses
---------- Forwarded message ----------
From: Anilesh K. Krishnaswamy
Date: Wed, Nov 8, 2017 at 11:07 AM
Subject: [RAIN] UPDATE: Joint GSB/RAIN seminar - Jon Kleinberg - today at
3:45 PM
To: internetalgs at lists.stanford.edu
Title: Inherent Trade-Offs in the Fair Determination of Risk Scores
Abstract: Recent discussion in the public sphere about algorithmic
classification has involved tension between competing notions of what it
means for a probabilistic classification to be fair to different groups. We
formalize three fairness conditions that lie at the heart of these debates,
and we prove that except in highly constrained special cases, there is no
method that can satisfy these three conditions simultaneously. Moreover,
even satisfying all three conditions approximately requires that the data
lie in an approximate version of one of the constrained special cases
identified by our theorem. These results suggest some of the ways in which
key notions of fairness are incompatible with each other, and hence provide
a framework for thinking about the trade-offs between them.
On Wed, Nov 8, 2017 at 10:35 AM, Anilesh K. Krishnaswamy <
anilesh at stanford.edu> wrote:
> Hi everyone,
>
> Jon Kleinberg from Cornell is speaking at a joint GSB/RAIN seminar today.
> Venue: GSB C105
> Time: 3:45 - 5 pm.
>
> See you there!
> --
> ------------------------------------------------------------
> ----------------
> Anilesh K. Krishnaswamy,
> Ph.D. candidate,
> Department of Electrical Engineering,
> Stanford University.
> Cell: 650-387-7272 <(650)%20387-7272>
>
--
----------------------------------------------------------------------------
Anilesh K. Krishnaswamy,
Ph.D. candidate,
Department of Electrical Engineering,
Stanford University.
Cell: 650-387-7272 <(650)%20387-7272>
--++**==--++**==--++**==--++**==--++**==--++**==--++**==
internetalgs mailing list
internetalgs at lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/internetalgs
--
You received this message because you are subscribed to the Google Groups
"stanfordml" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to stanfordml+unsubscribe at googlegroups.com.
To post to this group, send email to stanfordml at googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From silas at stanford.edu Thu Nov 9 10:54:21 2017
From: silas at stanford.edu (Shashwat Silas)
Date: Thu, 9 Nov 2017 18:54:21 +0000
Subject: [theory-seminar] Theory Seminar Today
Message-ID:
Hi all,
There will be a theory seminar today at 415pm in Gates 463A. See you there!
Clement Canonne (Stanford)
When Fourier SIIRVs: Fourier-Based Testing for Families of Distributions
We study the general problem of testing whether an unknown discrete probability distribution belongs to a specified family of distributions. More specifically, given a distribution family PP and sample access to an unknown discrete distribution pp, we want to distinguish (with high probability) between the case that pp is in PP and the case that pp is eps-far, in total variation distance, from every distribution in PP. This is the archetypal hypothesis testing problem that has received significant attention in statistics and, over the past decade and roughly a half, in theoretical computer science.
Of course, the sample complexity of this general inference task depends on the underlying family PP. The gold standard in distribution testing is to design sample-optimal and computationally efficient algorithms for this task, as a function of PP. The main contribution of this work is a simple and general testing technique that is applicable to /all/ distribution families, and is particularly suited to those whose /Fourier spectrum/ satisfies a certain approximate /sparsity/ property. To the best of our knowledge, ours is the first use of the Fourier transform in the context of distribution testing.
We apply our Fourier-based framework to obtain near sample-optimal and computationally efficient testers for the following fundamental distribution families: Sums of Independent Integer Random Variables (SIIRVs), Poisson Multinomial Distributions (PMDs), and Discrete Log-Concave Distributions. For the first two, ours are the first non-trivial testers in the literature, vastly generalizing previous work on testing Poisson Binomial Distributions. For the third, our tester improves on prior work in both sample and time complexity.
Joint work with Ilias Diakonikolas (USC) and Alistair Stewart (USC).
Thanks,
Shashwat
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Fri Nov 10 09:53:37 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Fri, 10 Nov 2017 09:53:37 -0800
Subject: [theory-seminar] Next TCS+ talk: Wednesday, November 15,
Vinod Vaikuntanathan, MIT
In-Reply-To:
References:
Message-ID: <9ae854c7-a890-e691-f024-1ac228133e2a@stanford.edu>
Hi everyone,
While TCS+ talks are usually every other week*, this month is different
due to Thanksgiving: and the next TCS+ talk will take place *this coming
Wednesday*, the 15th, at 10am in the conference room (463A).
Vinod Vaikuntanathan, from MIT, will tell us (remotely) about /"Program
Obfuscation and Random CSPs: The Love-Hate Relationship."/? (see below
for the abstract and details)
Since 10am can still be considered pre-dawn, for some definition of
dawn, there will be breakfast to entice you to come. (A bait generously
sponsored by Mary, and implemented by Avishay who is volunteering this
time. Worth showing up at 9:55am, I'd say.)
Hope to see you there,
-- Cl?ment
* biweekly? Fortnightly? Who knows.
https://www.merriam-webster.com/dictionary/biweekly
-------------------------------
Speaker: Vinod Vaikuntanathan (MIT)
Title: Program Obfuscation and Random CSPs: The Love-Hate Relationship
Abstract: A recent line of work shows how to construct
indistinguishability obfuscation under two assumptions: (a) that there
exist k-linear maps for some constant k; and (b) that certain random
O(k)-constraint satisfaction problems (CSPs) are hard in an appropriate
sense. The latest of these works (by Lin and Tessaro) assumes the
existence of 3-linear maps and the hardness of certain random 3-CSPs. We
have 1-linear maps since the 1970s; 2-linear maps since the 1990s; but
the existence of 3-linear maps is wide open. On the other hand, we do
have reasonable constructions of "secure" random 3-CSPs. The first part
of the talk will describe these developments.
Much more surprising was a result (from the same work of Lin and
Tessaro) which showed a construction from 2-linear maps and the hardness
of random 2-CSPs over a large alphabet. Overnight, the burden of
existence of IO went from the question of whether 3-linear maps exist to
the completely unrelated question of whether random 2-CSPs over large
alphabets is hard. In a nutshell, they require the existence of
pseudo-random generators G: \Sigma^n \to {0,1}^m for some poly(n)-size
alphabet \Sigma where each output bit depends on at most two input
alphabet symbols, and which achieve sufficiently large stretch. In the
second part of the talk, we will present a polynomial-time algorithm
that breaks these random CSPs.
Based on joint work with Alex Lombardi (MIT) and Rachel Lin (UCSB).
_______________________________________________
Tcsplus_announce mailing list
Tcsplus_announce at cs.nyu.edu
https://cs.nyu.edu/mailman/listinfo/tcsplus_announce
From moses at cs.stanford.edu Mon Nov 13 13:39:39 2017
From: moses at cs.stanford.edu (Moses Charikar)
Date: Mon, 13 Nov 2017 13:39:39 -0800
Subject: [theory-seminar] Motwani CS Theory Colloquium: Manuel Blum (Nov 16)
Message-ID:
The Motwani Distinguished Lectures are a series of theory colloquia aimed
at a broad audience. The next lecture in the series will be given
Thursday, Nov 16, by Manuel Blum from CMU, describing a computational model
for consciousness. It should be a great talk---you should definitely attend
if you can!
The talk is at 4:15 PM on Nov 16 in the Mackenzie Room (Huang
Engineering Center, Room 300,
http://campus-map.stanford.edu/?id=&lat=37.43413296203056&lng=-122.
17429865&zoom=15&srch=Huang%20Engineering%20Center).
There will be a reception with refreshments immediately following the talk.
Hope to see you there!
Moses
---------
TITLE:
Can a Machine be Conscious? Towards a Computational Model of
Consciousness.
ABSTRACT:
Thanks to major advances in neuroscience, we are on the brink of a
scientific understanding of how the brain achieves consciousness. This
talk will describe neuroscientist Bernard Baars's Global Workspace Model
(GWM) of the brain and propose a formal Turing-Machine-like computational
model inspired by it for understanding consciousness. One of several
consequences of this Model is the possibility of free will in a completely
deterministic world. Another deals with the possibility of building
machines that are conscious.
This talk is suitable for college students at all levels, engineers,
mathematicians, and anyone who has ever wondered about consciousness.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From hongyang at cs.stanford.edu Mon Nov 13 21:49:56 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Mon, 13 Nov 2017 21:49:56 -0800
Subject: [theory-seminar] [theory-lunch] Michael Kim on algorithmic fairness
Message-ID:
Hi Everyone
This week Michael will tell us about his recent work on *Calibration for
the (Computationally-Identifiable) Masses* (abstract follows below).
As before, we meet from 12:00pm to 1:00pm at Gates 463A.
==================================
Title: Calibration for the (Computationally-Identifiable) Masses
Abstract: As algorithms increasingly inform and influence decisions made
about individuals, it becomes increasingly important to address concerns
that these algorithms might be discriminatory. In this talk, we introduce
multicalbration ? a new measure of algorithmic fairness that aims to
mitigate concerns about discrimination that is introduced during the
process of learning a predictor from data. Multicalibration guarantees
accurate (calibrated) predictions for every subpopulation that can be
identified within a specified class of computations. We think of the class
as being quite rich; in particular, it can contain many and overlapping
subgroups of a protected group. We show that in many settings this strong
notion of protection from discrimination is both attainable and aligned
with the goal of obtaining accurate predictions. This talk will begin with
an overview of some of the existing work on fairness in algorithm design
and how multicalibration adds to the fairness toolkit. Then, we will
discuss our algorithm for learning a multicalibrated predictor, the
computational complexity of this task, and draw connections to
computational learning models such as agnostic learning.
==================================
Best,
Hongyang
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From hongyang at cs.stanford.edu Mon Nov 13 21:49:56 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Mon, 13 Nov 2017 21:49:56 -0800
Subject: [theory-seminar] [theory-lunch] Michael Kim on algorithmic fairness
Message-ID:
Hi Everyone
This week Michael will tell us about his recent work on *Calibration for
the (Computationally-Identifiable) Masses* (abstract follows below).
As before, we meet from 12:00pm to 1:00pm at Gates 463A.
==================================
Title: Calibration for the (Computationally-Identifiable) Masses
Abstract: As algorithms increasingly inform and influence decisions made
about individuals, it becomes increasingly important to address concerns
that these algorithms might be discriminatory. In this talk, we introduce
multicalbration ? a new measure of algorithmic fairness that aims to
mitigate concerns about discrimination that is introduced during the
process of learning a predictor from data. Multicalibration guarantees
accurate (calibrated) predictions for every subpopulation that can be
identified within a specified class of computations. We think of the class
as being quite rich; in particular, it can contain many and overlapping
subgroups of a protected group. We show that in many settings this strong
notion of protection from discrimination is both attainable and aligned
with the goal of obtaining accurate predictions. This talk will begin with
an overview of some of the existing work on fairness in algorithm design
and how multicalibration adds to the fairness toolkit. Then, we will
discuss our algorithm for learning a multicalibrated predictor, the
computational complexity of this task, and draw connections to
computational learning models such as agnostic learning.
==================================
Best,
Hongyang
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Tue Nov 14 06:08:32 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Tue, 14 Nov 2017 06:08:32 -0800
Subject: [theory-seminar] Next TCS+ talk: Wednesday, November 15,
Vinod Vaikuntanathan, MIT
In-Reply-To: <9ae854c7-a890-e691-f024-1ac228133e2a@stanford.edu>
References:
<9ae854c7-a890-e691-f024-1ac228133e2a@stanford.edu>
Message-ID:
Reminder: this is *tomorrow*, 10 am!
Cl?ment
On 11/10/2017 09:53 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> While TCS+ talks are usually every other week*, this month is different
> due to Thanksgiving: and the next TCS+ talk will take place *this coming
> Wednesday*, the 15th, at 10am in the conference room (463A).
>
> Vinod Vaikuntanathan, from MIT, will tell us (remotely) about /"Program
> Obfuscation and Random CSPs: The Love-Hate Relationship."/? (see below
> for the abstract and details)
>
> Since 10am can still be considered pre-dawn, for some definition of
> dawn, there will be breakfast to entice you to come. (A bait generously
> sponsored by Mary, and implemented by Avishay who is volunteering this
> time. Worth showing up at 9:55am, I'd say.)
>
> Hope to see you there,
>
> -- Cl?ment
>
> * biweekly? Fortnightly? Who knows.
> https://www.merriam-webster.com/dictionary/biweekly
>
> -------------------------------
> Speaker: Vinod Vaikuntanathan (MIT)
> Title: Program Obfuscation and Random CSPs: The Love-Hate Relationship
>
> Abstract: A recent line of work shows how to construct
> indistinguishability obfuscation under two assumptions: (a) that there
> exist k-linear maps for some constant k; and (b) that certain random
> O(k)-constraint satisfaction problems (CSPs) are hard in an appropriate
> sense. The latest of these works (by Lin and Tessaro) assumes the
> existence of 3-linear maps and the hardness of certain random 3-CSPs. We
> have 1-linear maps since the 1970s; 2-linear maps since the 1990s; but
> the existence of 3-linear maps is wide open. On the other hand, we do
> have reasonable constructions of "secure" random 3-CSPs. The first part
> of the talk will describe these developments.
>
> Much more surprising was a result (from the same work of Lin and
> Tessaro) which showed a construction from 2-linear maps and the hardness
> of random 2-CSPs over a large alphabet. Overnight, the burden of
> existence of IO went from the question of whether 3-linear maps exist to
> the completely unrelated question of whether random 2-CSPs over large
> alphabets is hard. In a nutshell, they require the existence of
> pseudo-random generators G: \Sigma^n \to {0,1}^m for some poly(n)-size
> alphabet \Sigma where each output bit depends on at most two input
> alphabet symbols, and which achieve sufficiently large stretch. In the
> second part of the talk, we will present a polynomial-time algorithm
> that breaks these random CSPs.
>
> Based on joint work with Alex Lombardi (MIT) and Rachel Lin (UCSB).
>
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From moses at cs.stanford.edu Thu Nov 16 14:02:55 2017
From: moses at cs.stanford.edu (Moses Charikar)
Date: Thu, 16 Nov 2017 14:02:55 -0800
Subject: [theory-seminar] Fwd: Motwani CS Theory Colloquium: Manuel Blum
(Nov 16)
In-Reply-To:
References:
Message-ID:
Reminder - in 2 hours.
Hope to see all of you there! It would be great to have a big audience for
this special occasion.
Cheers,
Moses
---------- Forwarded message ----------
From: Moses Charikar
Date: Mon, Nov 13, 2017 at 1:39 PM
Subject: Motwani CS Theory Colloquium: Manuel Blum (Nov 16)
To: cstheory-special at mailman.stanford.edu, phd at cs.stanford.edu,
bs at cs.stanford.edu, ms at cs.stanford.edu, theory-seminar at lists.stanford.edu,
CS Faculty
The Motwani Distinguished Lectures are a series of theory colloquia aimed
at a broad audience. The next lecture in the series will be given
Thursday, Nov 16, by Manuel Blum from CMU, describing a computational model
for consciousness. It should be a great talk---you should definitely attend
if you can!
The talk is at 4:15 PM on Nov 16 in the Mackenzie Room (Huang
Engineering Center, Room 300,
http://campus-map.stanford.edu/?id=&lat=37.43413296203056&l
ng=-122.17429865&zoom=15&srch=Huang%20Engineering%20Center).
There will be a reception with refreshments immediately following the talk.
Hope to see you there!
Moses
---------
TITLE:
Can a Machine be Conscious? Towards a Computational Model of
Consciousness.
ABSTRACT:
Thanks to major advances in neuroscience, we are on the brink of a
scientific understanding of how the brain achieves consciousness. This
talk will describe neuroscientist Bernard Baars's Global Workspace Model
(GWM) of the brain and propose a formal Turing-Machine-like computational
model inspired by it for understanding consciousness. One of several
consequences of this Model is the possibility of free will in a completely
deterministic world. Another deals with the possibility of building
machines that are conscious.
This talk is suitable for college students at all levels, engineers,
mathematicians, and anyone who has ever wondered about consciousness.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Wed Nov 22 09:27:01 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 22 Nov 2017 09:27:01 -0800
Subject: [theory-seminar] TCS+ talk: Wednesday, November 29, Jon Kelner, MIT
In-Reply-To: <9fc0e7e4-8274-79bc-8088-ce3fb3237ac9@cms.caltech.edu>
References: <9fc0e7e4-8274-79bc-8088-ce3fb3237ac9@cms.caltech.edu>
Message-ID:
Hi everyone,
The next TCS+ talk will be a bit special (see announcement, title, and
abstract below), and dedicated to Michael Cohen, a grad student at MIT
who passed away recently.
Jon Kelner will be speaking about a work he had with Michael (and
others: Rasmus Kyng, John Peebles, Richard Peng, Anup Rao, Aaron
Sidford, and Adrian Vladu), "Almost-Linear-Time Algorithms for Markov
Chains and New Spectral Primitives for Directed Graphs."
This will be next Wednesday, in Gates 465, at 10am as usual (with
breakfast at 9:55).
Hope to see you there,
-- Cl?ment
-------- Forwarded Message --------
Many of you will be aware of the recent passing away of MIT graduate
student Michael Cohen. At TCS+ we had asked Michael to give a talk;
although he responded positively, he had also told us he'd want to wait
for a really cool result to tell us all about.
The next TCS+ talk is dedicated to Michael's memory. The talk will be
given by Jon Kelner, from MIT, on work in which Michael played an
important part (title and abstract below).
The talk will take place this coming Wednesday, November 29th at 1:00 PM
Eastern Time (10:00 AM Pacific Time, 19:00 Central European Time, 18:00
UTC). Jon Kelner will speak about "Almost-Linear-Time Algorithms for
Markov Chains and New Spectral Primitives for Directed Graphs".
Please sign up on the online form at
https://sites.google.com/site/plustcs/livetalk/live-seat-reservation if
you wish to join the talk as a group.
Hoping to see you all there,
The organizers
-------------------------------
Speaker: Jon Kelner (MIT)
Title: Almost-Linear-Time Algorithms for Markov Chains and New Spectral
Primitives for Directed Graphs
Abstract: In the analysis of Markov chains, there has been a
longstanding algorithmic gap between the general case, corresponding to
random walks on directed graphs, and the special case of reversible
chains, for which the corresponding graph can be taken to be undirected.
This begins with the most basic computational task, computing the
stationary distribution, and persists for many of the fundamental
problems associated with random walks, such as computing hitting and
commute times, escape probabilities, and personalized PageRank vectors.
In the undirected case, there are algorithms for all of these problems
that run in linear or nearly-linear time, whereas it was unknown in the
directed case whether one could solve any of them more efficiently than
an arbitrary linear system.
More broadly, this gap has its origins in a substantial discrepancy
between the state of algorithmic spectral graph theory in the undirected
and directed settings. While the undirected case has a richly developed
theory and a powerful collection of algorithmic tools, similar results
have remained elusive for directed graphs.
In this talk, I will begin to address this by giving an algorithmic
framework that solves all of the problems listed above in almost-linear
time. To do so, I will develop the first directed versions of several
foundational primitives from undirected algorithmic spectral graph
theory that had not been known to exist for directed graphs, notably
including the first directed version of graph sparsification and an
almost-linear-time solver for directed Laplacian systems. If time
permits, I will also briefly discuss more recent work that improves the
running time to be nearly linear, thereby eliminating the gap between
the undirected and directed versions of these problems (up to
polylogarithmic factors).
This talk is based on work with Michael Cohen, Rasmus Kyng, John
Peebles, Richard Peng, Anup Rao, Aaron Sidford, and Adrian Vladu.
From marykw at stanford.edu Fri Nov 24 12:08:31 2017
From: marykw at stanford.edu (Mary Wootters)
Date: Fri, 24 Nov 2017 12:08:31 -0800
Subject: [theory-seminar] BONUS theory seminar! Jelani Nelson, Tuesday 11/28.
Message-ID:
Hi all,
This week we will have a **bonus** theory seminar on Tuesday 11/28 at
1:30pm (note the non-standard time) by Jelani Nelson, in Gates 463A. Title
and abstract below. Jelani will be around for the rest of the day, so let
me know if you'd like to meet with him.
--Mary
Jelani Nelson (Harvard)
Tuesday 11/28, 1:30pm, Gates 463A
Abstract:
Dimensionality reduction in Euclidean space, as attainable by the
Johnson-Lindenstrauss lemma, has been a fundamental tool in algorithm
design and machine learning. The JL lemma states that any n point
subset of l_2 can be mapped to l_2^m with distortion 1+epsilon, where
m = O((log n) / epsilon^2). In this talk, I discuss our recent proof
that the JL lemma is optimal, in the sense that for any n, d, epsilon,
where epsilon is not too small, there is a point set X in l_2^d such
that any f:X->l_2^m with 1+epsilon must have m = Omega(epsilon^{-2}
log n). I will also discuss some subsequent work and future
directions.
Joint work with Kasper Green Larsen (Aarhus University).
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From clc2200 at columbia.edu Sun Nov 26 17:17:50 2017
From: clc2200 at columbia.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Sun, 26 Nov 2017 17:17:50 -0800
Subject: [theory-seminar] Theory Happy Hour: S01E00 (Pilot)
Message-ID: <1cf1784a-947f-a129-0e61-67fde751c4fe@cs.columbia.edu>
Hi everyone,
As there does not seem to be (yet) a regular series of happy hour events
for the Theory group, here's an attempt at mitigating this unfortunate
state of affairs.
*tl;dr:* December 7th, 6pm, @ the Treehouse
(https://goo.gl/maps/wAzimZavr2T2). Details deferred to the full version.
-----------------------
Come and have a drink (and finger food) with your fellow grad students,
faculty, postdocs at the Treehouse on Thursday, December 7th.
Departure from the Gates building, 4th floor, at 5h41.
The occasion to mingle and talk with the fellow people mentioned above.
Event sponsored by the faculty (Mary and Greg, within reasonable
bounds): and?hopefully? the first of a long series.
Best,
--
Cl?ment
From hongyang at cs.stanford.edu Mon Nov 27 19:02:35 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Mon, 27 Nov 2017 19:02:35 -0800
Subject: [theory-seminar] [theory-lunch] Warut Suksompong on resource
allocation
Message-ID:
Hi Everyone
This Thursday noon, Warut Suksompong will tell us about "Assigning a Small
Agreeable Set of Indivisible Items to Multiple Agents" (abstract follows
below).
As before, we meet from 12:00pm to 1:00pm at Gates 463A.
================================
Title: Assigning a Small Agreeable Set of Indivisible Items to Multiple
Agents
Abstract: We study the problem of assigning a small subset of indivisible
items to a group of agents so that the subset is agreeable to all agents,
meaning that all agents value the subset as least as much as its
complement. For an arbitrary number of agents and items, we derive a tight
worst-case bound on the number of items that may need to be included in
such a subset. We then present polynomial-time algorithms that find an
agreeable subset whose size matches the worst-case bound when there are two
or three agents. Furthermore, we investigate the problem of efficiently
computing an agreeable subset whose size approximates the size of the
smallest agreeable subset for any given instance. We consider three
well-known models for representing the preferences of the agents---ordinal
preferences on single items, the value oracle model, and additive
utilities---and establish tight bounds on the approximation ratio that can
be obtained by algorithms running in polynomial time in each of these
models.
===============================
Best,
Hongyang
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From hongyang at cs.stanford.edu Mon Nov 27 19:02:35 2017
From: hongyang at cs.stanford.edu (Hongyang Zhang)
Date: Mon, 27 Nov 2017 19:02:35 -0800
Subject: [theory-seminar] [theory-lunch] Warut Suksompong on resource
allocation
Message-ID:
Hi Everyone
This Thursday noon, Warut Suksompong will tell us about "Assigning a Small
Agreeable Set of Indivisible Items to Multiple Agents" (abstract follows
below).
As before, we meet from 12:00pm to 1:00pm at Gates 463A.
================================
Title: Assigning a Small Agreeable Set of Indivisible Items to Multiple
Agents
Abstract: We study the problem of assigning a small subset of indivisible
items to a group of agents so that the subset is agreeable to all agents,
meaning that all agents value the subset as least as much as its
complement. For an arbitrary number of agents and items, we derive a tight
worst-case bound on the number of items that may need to be included in
such a subset. We then present polynomial-time algorithms that find an
agreeable subset whose size matches the worst-case bound when there are two
or three agents. Furthermore, we investigate the problem of efficiently
computing an agreeable subset whose size approximates the size of the
smallest agreeable subset for any given instance. We consider three
well-known models for representing the preferences of the agents---ordinal
preferences on single items, the value oracle model, and additive
utilities---and establish tight bounds on the approximation ratio that can
be obtained by algorithms running in polynomial time in each of these
models.
===============================
Best,
Hongyang
--
Hongyang Zhang ?????
cs.stanford.edu/people/hongyang/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From marykw at stanford.edu Tue Nov 28 08:27:24 2017
From: marykw at stanford.edu (Mary Wootters)
Date: Tue, 28 Nov 2017 08:27:24 -0800
Subject: [theory-seminar] Bonus theory seminar today: Jelani Nelson, 1:30pm
Message-ID:
Hi all,
Just a reminder about the bonus theory seminar today, 1:30pm in Gates 463A.
Jelani Nelson will be speaking. (I just realized I forgot the title in
the previous email! It's "Optimality of the Johnson-Lindenstrauss lemma."
Abstract below.)
Hope to see you all there!
Mary
---------- Forwarded message ----------
From: Mary Wootters
Date: Fri, Nov 24, 2017 at 12:08 PM
Subject: BONUS theory seminar! Jelani Nelson, Tuesday 11/28.
To: theory-seminar at lists.stanford.edu
Hi all,
This week we will have a **bonus** theory seminar on Tuesday 11/28 at
1:30pm (note the non-standard time) by Jelani Nelson, in Gates 463A. Title
and abstract below. Jelani will be around for the rest of the day, so let
me know if you'd like to meet with him.
--Mary
Jelani Nelson (Harvard)
Tuesday 11/28, 1:30pm, Gates 463A
Abstract:
Dimensionality reduction in Euclidean space, as attainable by the
Johnson-Lindenstrauss lemma, has been a fundamental tool in algorithm
design and machine learning. The JL lemma states that any n point
subset of l_2 can be mapped to l_2^m with distortion 1+epsilon, where
m = O((log n) / epsilon^2). In this talk, I discuss our recent proof
that the JL lemma is optimal, in the sense that for any n, d, epsilon,
where epsilon is not too small, there is a point set X in l_2^d such
that any f:X->l_2^m with 1+epsilon must have m = Omega(epsilon^{-2}
log n). I will also discuss some subsequent work and future
directions.
Joint work with Kasper Green Larsen (Aarhus University).
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Tue Nov 28 15:48:49 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Tue, 28 Nov 2017 15:48:49 -0800
Subject: [theory-seminar] TCS+ talk: Wednesday, November 29, Jon Kelner,
MIT
In-Reply-To:
References: <9fc0e7e4-8274-79bc-8088-ce3fb3237ac9@cms.caltech.edu>
Message-ID:
Reminder: this is tomorrow, 10am (with bagels at 9:55am)!
Cl?ment
On 11/22/2017 09:27 AM, Cl?ment Canonne wrote:
> Hi everyone,
>
> The next TCS+ talk will be a bit special (see announcement, title, and
> abstract below), and dedicated to Michael Cohen, a grad student at MIT
> who passed away recently.
>
> Jon Kelner will be speaking about a work he had with Michael (and
> others:? Rasmus Kyng, John Peebles, Richard Peng, Anup Rao, Aaron
> Sidford, and Adrian Vladu), "Almost-Linear-Time Algorithms for Markov
> Chains and New Spectral Primitives for Directed Graphs."
>
> This will be next Wednesday, in Gates 465, at 10am as usual (with
> breakfast at 9:55).
>
> Hope to see you there,
>
> -- Cl?ment
>
>
> -------- Forwarded Message --------
>
> Many of you will be aware of the recent passing away of MIT graduate
> student Michael Cohen. At TCS+ we had asked Michael to give a talk;
> although he responded positively, he had also told us he'd want to wait
> for a really cool result to tell us all about.
>
> The next TCS+ talk is dedicated to Michael's memory. The talk will be
> given by Jon Kelner, from MIT, on work in which Michael played an
> important part (title and abstract below).
>
> The talk will take place this coming Wednesday, November 29th at 1:00 PM
> Eastern Time (10:00 AM Pacific Time, 19:00 Central European Time, 18:00
> UTC). Jon Kelner will speak about "Almost-Linear-Time Algorithms for
> Markov Chains and New Spectral Primitives for Directed Graphs".
>
> Please sign up on the online form at
> https://sites.google.com/site/plustcs/livetalk/live-seat-reservation if
> you wish to join the talk as a group.
>
> Hoping to see you all there,
>
> The organizers
>
> -------------------------------
> Speaker: Jon Kelner (MIT)
> Title: Almost-Linear-Time Algorithms for Markov Chains and New Spectral
> Primitives for Directed Graphs
>
> Abstract: In the analysis of Markov chains, there has been a
> longstanding algorithmic gap between the general case, corresponding to
> random walks on directed graphs, and the special case of reversible
> chains, for which the corresponding graph can be taken to be undirected.
> ?This begins with the most basic computational task, computing the
> stationary distribution, and persists for many of the fundamental
> problems associated with random walks, such as computing hitting and
> commute times, escape probabilities, and personalized PageRank vectors.
> In the undirected case, there are algorithms for all of these problems
> that run in linear or nearly-linear time, whereas it was unknown in the
> directed case whether one could solve any of them more efficiently than
> an arbitrary linear system.
>
> More broadly, this gap has its origins in a substantial discrepancy
> between the state of algorithmic spectral graph theory in the undirected
> and directed settings.? While the undirected case has a richly developed
> theory and a powerful collection of algorithmic tools, similar results
> have remained elusive for directed graphs.
>
> In this talk, I will begin to address this by giving an algorithmic
> framework that solves all of the problems listed above in almost-linear
> time.? To do so, I will develop the first directed versions of several
> foundational primitives from undirected algorithmic spectral graph
> theory that had not been known to exist for directed graphs, notably
> including the first directed version of graph sparsification and an
> almost-linear-time solver for directed Laplacian systems.? If time
> permits, I will also briefly discuss more recent work that improves the
> running time to be nearly linear, thereby eliminating the gap between
> the undirected and directed versions of these problems (up to
> polylogarithmic factors).
>
> This talk is based on work with Michael Cohen, Rasmus Kyng, John
> Peebles, Richard Peng, Anup Rao, Aaron Sidford, and Adrian Vladu.
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From silas at stanford.edu Wed Nov 29 16:19:35 2017
From: silas at stanford.edu (Shashwat Silas)
Date: Thu, 30 Nov 2017 00:19:35 +0000
Subject: [theory-seminar] Theory Seminar Tomorrow
Message-ID:
Hi all,
As usual, there will be a theory seminar tomorrow in Gates 463A at 415pm. Hope you can make it!
Jakub Tarnawski (EPFL)
A Constant-Factor Approximation Algorithm for the Asymmetric Traveling Salesman Problem
We give a constant-factor approximation algorithm for the asymmetric traveling salesman problem. Our approximation guarantee is analyzed with respect to the standard LP relaxation, and thus our result confirms the conjectured constant integrality gap of that relaxation. Our techniques build upon the constant-factor approximation algorithm for the special case of node-weighted metrics. Specifically, we give a generic reduction to structured instances that resemble but are more general than those arising from node-weighted metrics. For those instances, we then solve Local-Connectivity ATSP, a problem known to be equivalent (in terms of constant-factor approximation) to the asymmetric traveling salesman problem. This is joint work with Ola Svensson and L?szl? V?gh.
Thanks,
Shashwat
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From silas at stanford.edu Thu Nov 30 11:23:13 2017
From: silas at stanford.edu (Shashwat Silas)
Date: Thu, 30 Nov 2017 19:23:13 +0000
Subject: [theory-seminar] Theory Seminar Today
Message-ID:
Hi all,
Just a reminder that theory seminar will be today at 415PM in Gates 463A.
Jakub Tarnawski (EPFL)
A Constant-Factor Approximation Algorithm for the Asymmetric Traveling Salesman Problem
We give a constant-factor approximation algorithm for the asymmetric traveling salesman problem. Our approximation guarantee is analyzed with respect to the standard LP relaxation, and thus our result confirms the conjectured constant integrality gap of that relaxation. Our techniques build upon the constant-factor approximation algorithm for the special case of node-weighted metrics. Specifically, we give a generic reduction to structured instances that resemble but are more general than those arising from node-weighted metrics. For those instances, we then solve Local-Connectivity ATSP, a problem known to be equivalent (in terms of constant-factor approximation) to the asymmetric traveling salesman problem. This is joint work with Ola Svensson and L?szl? V?gh.
Thanks,
Shashwat
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at stanford.edu Thu Nov 30 13:02:07 2017
From: ccanonne at stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Thu, 30 Nov 2017 13:02:07 -0800
Subject: [theory-seminar] Theory Happy Hour: S01E00 (Pilot)
In-Reply-To: <1cf1784a-947f-a129-0e61-67fde751c4fe@cs.columbia.edu>
References: <1cf1784a-947f-a129-0e61-67fde751c4fe@cs.columbia.edu>
Message-ID: <81bc51e8-09c2-e412-bb55-3bc5e61d8287@stanford.edu>
Just a reminder (since I failed to announce it during the theory lunch):
this is next Thursday!
Cl?ment
On 11/26/2017 05:17 PM, Cl?ment Canonne wrote:
> Hi everyone,
>
> As there does not seem to be (yet) a regular series of happy hour events
> for the Theory group, here's an attempt at mitigating this unfortunate
> state of affairs.
>
> *tl;dr:* December 7th, 6pm, @ the Treehouse
> (https://goo.gl/maps/wAzimZavr2T2). Details deferred to the full version.
>
> -----------------------
>
> Come and have a drink (and finger food) with your fellow grad students,
> faculty, postdocs at the Treehouse on Thursday, December 7th.
> Departure from the Gates building, 4th floor, at 5h41.
> The occasion to mingle and talk with the fellow people mentioned above.
> Event sponsored by the faculty (Mary and Greg, within reasonable
> bounds): and?hopefully? the first of a long series.
>
>
> Best,