planted cliques, iterative thresholding and message ...yash/slides/simons.pdf · deshpande,...
TRANSCRIPT
![Page 1: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/1.jpg)
Planted Cliques, Iterative Thresholding and MessagePassing Algorithms
Yash Deshpande and Andrea Montanari
Stanford University
November 5, 2013
Deshpande, Montanari Planted Cliques November 5, 2013 1 / 49
![Page 2: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/2.jpg)
Problem Definition
Given distributions Q0;Q1,
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S
Q0 otherwise.
Aij = Aji
Problem: Given A, identify S
Deshpande, Montanari Planted Cliques November 5, 2013 2 / 49
![Page 3: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/3.jpg)
Problem Definition
Given distributions Q0;Q1,
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S
Q0 otherwise.
Aij = Aji
S
S
A =
Q1
Q0
Problem: Given A, identify S
Deshpande, Montanari Planted Cliques November 5, 2013 2 / 49
![Page 4: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/4.jpg)
Problem Definition
Given distributions Q0;Q1,
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S
Q0 otherwise.
Aij = Aji
S
S
A =
Q1
Q0
Problem: Given A, identify S
Deshpande, Montanari Planted Cliques November 5, 2013 2 / 49
![Page 5: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/5.jpg)
An Example
Q1 = N(�; 1)
Q0 = N(0; 1):
S
S
A =
Q1
Q0
Deshpande, Montanari Planted Cliques November 5, 2013 3 / 49
![Page 6: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/6.jpg)
An Example
A = �uSuTS + Z
λ
λ
λ
λ0
00
∼ N(0, 1)
Data = Sparse, Low Rank + noise
Deshpande, Montanari Planted Cliques November 5, 2013 4 / 49
![Page 7: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/7.jpg)
An Example
A = �uSuTS + Z
λ
λ
λ
λ0
00
∼ N(0, 1)
Data = Sparse, Low Rank + noise
Deshpande, Montanari Planted Cliques November 5, 2013 4 / 49
![Page 8: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/8.jpg)
Much work in statistics
Denoising:
y = x + noise
[Donoho, Jin 2004], [Arias-Castro, Candes, Durand 2011][Arias-Castro, Bubeck, Lugosi 2012]
Sparse signal recovery:
y = Ax + noise
[Candes, Romberg, Tao 2004], [Chen, Donoho 1998]
The simple example combines different structures
Deshpande, Montanari Planted Cliques November 5, 2013 5 / 49
![Page 9: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/9.jpg)
Much work in statistics
Denoising:
y = x + noise
[Donoho, Jin 2004], [Arias-Castro, Candes, Durand 2011][Arias-Castro, Bubeck, Lugosi 2012]
Sparse signal recovery:
y = Ax + noise
[Candes, Romberg, Tao 2004], [Chen, Donoho 1998]
The simple example combines different structures
Deshpande, Montanari Planted Cliques November 5, 2013 5 / 49
![Page 10: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/10.jpg)
Much work in statistics
Denoising:
y = x + noise
[Donoho, Jin 2004], [Arias-Castro, Candes, Durand 2011][Arias-Castro, Bubeck, Lugosi 2012]
Sparse signal recovery:
y = Ax + noise
[Candes, Romberg, Tao 2004], [Chen, Donoho 1998]
The simple example combines different structures
Deshpande, Montanari Planted Cliques November 5, 2013 5 / 49
![Page 11: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/11.jpg)
Our Running Example: Planted Cliques
A is “adjacency” matrix
Q1 = �+1
Q0 =12�+1 +
12��1:
S forms a clique in the graph
[Jerrum, 1992]
Deshpande, Montanari Planted Cliques November 5, 2013 6 / 49
![Page 12: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/12.jpg)
Our Running Example: Planted Cliques
A is “adjacency” matrix
Q1 = �+1
Q0 =12�+1 +
12��1:
S forms a clique in the graph
[Jerrum, 1992]
Deshpande, Montanari Planted Cliques November 5, 2013 6 / 49
![Page 13: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/13.jpg)
Our Running Example: Planted Cliques
I Average case version ofMAX-CLIQUE
I Communities in networks
Deshpande, Montanari Planted Cliques November 5, 2013 7 / 49
![Page 14: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/14.jpg)
Our Running Example: Planted Cliques
I Average case version ofMAX-CLIQUE
I Communities in networks
Deshpande, Montanari Planted Cliques November 5, 2013 7 / 49
![Page 15: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/15.jpg)
How large should S be?
Let jS j = k
Size of largest clique in G�n; 1
2
�Second moment calculation ) k > 2 log2 n.
Deshpande, Montanari Planted Cliques November 5, 2013 8 / 49
![Page 16: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/16.jpg)
How large should S be?
Let jS j = k
Size of largest clique in G�n; 1
2
�Second moment calculation ) k > 2 log2 n.
Deshpande, Montanari Planted Cliques November 5, 2013 8 / 49
![Page 17: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/17.jpg)
Progress(0)
Complexity
k
n2 logn
2 log2 n
Exhaustive search
Deshpande, Montanari Planted Cliques November 5, 2013 9 / 49
![Page 18: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/18.jpg)
A Naive Algorithm
Pick k largest degree vertices of G as bS .
Deshpande, Montanari Planted Cliques November 5, 2013 10 / 49
![Page 19: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/19.jpg)
A Naive Algorithm
Pick k largest degree vertices of G as bS .
Deshpande, Montanari Planted Cliques November 5, 2013 10 / 49
![Page 20: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/20.jpg)
Analysis of NAIVE
If i =2 S :
deg(i) = Binomial�n � 1;
12
�) max
i =2Sdeg(i) � n
2+ O(
qn log n)
If i 2 S :
deg(i) = k � 1 + Binomial�n � k + 1;
12
�) min
i2Sdeg(i) � k � 1
2+
n
2� O(
qn log n)
NAIVE works if: k � O(pn log n)
Deshpande, Montanari Planted Cliques November 5, 2013 11 / 49
![Page 21: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/21.jpg)
Analysis of NAIVE
If i =2 S :
deg(i) = Binomial�n � 1;
12
�) max
i =2Sdeg(i) � n
2+ O(
qn log n)
If i 2 S :
deg(i) = k � 1 + Binomial�n � k + 1;
12
�) min
i2Sdeg(i) � k � 1
2+
n
2� O(
qn log n)
NAIVE works if: k � O(pn log n)
Deshpande, Montanari Planted Cliques November 5, 2013 11 / 49
![Page 22: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/22.jpg)
Analysis of NAIVE
If i =2 S :
deg(i) = Binomial�n � 1;
12
�) max
i =2Sdeg(i) � n
2+ O(
qn log n)
If i 2 S :
deg(i) = k � 1 + Binomial�n � k + 1;
12
�) min
i2Sdeg(i) � k � 1
2+
n
2� O(
qn log n)
NAIVE works if: k � O(pn log n)
Deshpande, Montanari Planted Cliques November 5, 2013 11 / 49
![Page 23: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/23.jpg)
Analysis of NAIVE
If i =2 S :
deg(i) = Binomial�n � 1;
12
�) max
i =2Sdeg(i) � n
2+ O(
qn log n)
If i 2 S :
deg(i) = k � 1 + Binomial�n � k + 1;
12
�) min
i2Sdeg(i) � k � 1
2+
n
2� O(
qn log n)
NAIVE works if: k � O(pn log n)
Deshpande, Montanari Planted Cliques November 5, 2013 11 / 49
![Page 24: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/24.jpg)
Progress(1)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
[Kučera, 1995]
Deshpande, Montanari Planted Cliques November 5, 2013 12 / 49
![Page 25: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/25.jpg)
Spectral Method
A = uSuTS + Z
1
1
1
10
00
0
±1
±1±1
Hopefully v1(A) � uS
Deshpande, Montanari Planted Cliques November 5, 2013 13 / 49
![Page 26: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/26.jpg)
Spectral Method
A = uSuTS + Z
1
1
1
10
00
0
±1
±1±1
Hopefully v1(A) � uS
Deshpande, Montanari Planted Cliques November 5, 2013 13 / 49
![Page 27: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/27.jpg)
Analysis of SPECTRAL
1pnA =
�kpn
�| {z }
�
eSeTS +
Zpn:
By standard linear algebra:
�� Zp
n
2� 2 =) hv1(A); eSi � 1� �(�)
SPECTRAL works if k � Cpn.
Deshpande, Montanari Planted Cliques November 5, 2013 14 / 49
![Page 28: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/28.jpg)
Progress(2)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
n2 log n
C√n
[Alon, Krivelevich and Sudakov, 1998]
Deshpande, Montanari Planted Cliques November 5, 2013 15 / 49
![Page 29: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/29.jpg)
Progress(2)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
n2 log n
C√n
[Ames, Vavasis 2009], [Feige, Krauthgamer 2000]
Deshpande, Montanari Planted Cliques November 5, 2013 16 / 49
![Page 30: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/30.jpg)
Progress(3)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
n2 log n
C√n1.65
√n
[Dekel, Gurel-Gurevich and Peres, 2010]
Deshpande, Montanari Planted Cliques November 5, 2013 17 / 49
![Page 31: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/31.jpg)
Progress(3)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
n2 log n
C√n1.261
√n
[Dekel, Gurel-Gurevich and Peres, 2010]
Deshpande, Montanari Planted Cliques November 5, 2013 18 / 49
![Page 32: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/32.jpg)
Progress(3)
Complexity
k
n2 logn
2 log2 n
n2
√n log n
n2 log n
C√n1.261
√n
nr+2
C√
n2r
[Alon, Krivelevich and Sudakov, 1998]
Deshpande, Montanari Planted Cliques November 5, 2013 19 / 49
![Page 33: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/33.jpg)
Phase transition in spectrum of A
Deshpande, Montanari Planted Cliques November 5, 2013 20 / 49
![Page 34: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/34.jpg)
Phase transition in spectrum of A
If k � (1 + ")pn
2−2
Limiting Spectral Density
hv1(A); eSi � �(")
Deshpande, Montanari Planted Cliques November 5, 2013 20 / 49
![Page 35: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/35.jpg)
Phase transition in spectrum of A
If k � (1 + ")pn
2−2
Limiting Spectral Density
hv1(A); eSi � �(")
If k � (1� ")pn
2−2
Limiting Spectral Density
hv1(A); eSi � n�1=2+�0(")
Deshpande, Montanari Planted Cliques November 5, 2013 20 / 49
![Page 36: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/36.jpg)
Phase transition in spectrum of A
If k � (1 + ")pn
2−2
Limiting Spectral Density
hv1(A); eSi � �(")
If k � (1� ")pn
2−2
Limiting Spectral Density
hv1(A); eSi � n�1=2+�0(")
[Knowles, Yin, 2011]
Deshpande, Montanari Planted Cliques November 5, 2013 20 / 49
![Page 37: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/37.jpg)
Seems much harder than it looks!
I “Statistical algorithms” fail if k = n1=2��: [Feldman et al., 2012]
I r -Lovász-Schrijver fails for k � pn=2r : [Feige, Krauthgamer, 2002]
I r -Lasserre fails for k �pn
(log n)r2: [Widgerson and Meka, 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49
![Page 38: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/38.jpg)
Seems much harder than it looks!
I “Statistical algorithms” fail if k = n1=2��: [Feldman et al., 2012]
I r -Lovász-Schrijver fails for k � pn=2r : [Feige, Krauthgamer, 2002]
I r -Lasserre fails for k �pn
(log n)r2: [Widgerson and Meka, 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49
![Page 39: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/39.jpg)
Seems much harder than it looks!
I “Statistical algorithms” fail if k = n1=2��: [Feldman et al., 2012]
I r -Lovász-Schrijver fails for k � pn=2r : [Feige, Krauthgamer, 2002]
I r -Lasserre fails for k �pn
(log n)r2: [Widgerson and Meka, 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49
![Page 40: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/40.jpg)
Our result
Theorem (Deshpande, Montanari, 2013)
If jS j = k � (1 + ")pn=e, there exists an O(n2 log n) time algorithm that
identifies S with high probability.
I will present:1 A (wrong) heuristic analysis2 How to fix the heuristic3 Lower bounds
Deshpande, Montanari Planted Cliques November 5, 2013 22 / 49
![Page 41: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/41.jpg)
Our result
Theorem (Deshpande, Montanari, 2013)
If jS j = k � (1 + ")pn=e, there exists an O(n2 log n) time algorithm that
identifies S with high probability.
I will present:1 A (wrong) heuristic analysis2 How to fix the heuristic3 Lower bounds
Deshpande, Montanari Planted Cliques November 5, 2013 22 / 49
![Page 42: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/42.jpg)
Our result
Theorem (Deshpande, Montanari, 2013)
If jS j = k � (1 + ")pn=e, there exists an O(n2 log n) time algorithm that
identifies S with high probability.
I will present:1 A (wrong) heuristic analysis2 How to fix the heuristic3 Lower bounds
Deshpande, Montanari Planted Cliques November 5, 2013 22 / 49
![Page 43: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/43.jpg)
Iterative Thresholding
The power iteration:
v t+1 = A v t :
Improvement:
v t+1 = AFt(vt):
where Ft(v) = (ft(v1); ft(v2); � � � ; ft(vn))T
Choose ft(�) to exploit sparsity of eS
Deshpande, Montanari Planted Cliques November 5, 2013 23 / 49
![Page 44: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/44.jpg)
Iterative Thresholding
The power iteration:
v t+1 = A v t :
Improvement:
v t+1 = AFt(vt):
where Ft(v) = (ft(v1); ft(v2); � � � ; ft(vn))T
Choose ft(�) to exploit sparsity of eS
Deshpande, Montanari Planted Cliques November 5, 2013 23 / 49
![Page 45: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/45.jpg)
Iterative Thresholding
The power iteration:
v t+1 = A v t :
Improvement:
v t+1 = AFt(vt):
where Ft(v) = (ft(v1); ft(v2); � � � ; ft(vn))T
Choose ft(�) to exploit sparsity of eS
Deshpande, Montanari Planted Cliques November 5, 2013 23 / 49
![Page 46: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/46.jpg)
Iterative Thresholding
The power iteration:
v t+1 = A v t :
Improvement:
v t+1 = AFt(vt):
where Ft(v) = (ft(v1); ft(v2); � � � ; ft(vn))T
Choose ft(�) to exploit sparsity of eS
Deshpande, Montanari Planted Cliques November 5, 2013 23 / 49
![Page 47: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/47.jpg)
Iterative Thresholding
The power iteration:
v t+1 = A v t :
Improvement:
v t+1 = AFt(vt):
where Ft(v) = (ft(v1); ft(v2); � � � ; ft(vn))T
Choose ft(�) to exploit sparsity of eS
Deshpande, Montanari Planted Cliques November 5, 2013 23 / 49
![Page 48: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/48.jpg)
(Wrong) Analysis
v t+1i =
1pn
Xj
Aij ft(vtj ):
Aij are random �1 r.v.
Use Central Limit Theorem for v t+1i
Deshpande, Montanari Planted Cliques November 5, 2013 24 / 49
![Page 49: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/49.jpg)
(Wrong) Analysis
v t+1i =
1pn
Xj
Aij ft(vtj ):
Aij are random �1 r.v.
Use Central Limit Theorem for v t+1i
Deshpande, Montanari Planted Cliques November 5, 2013 24 / 49
![Page 50: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/50.jpg)
(Wrong) Analysis
If i =2 S :
v t+1i =
1pn
Xj
Aij ft(vtj )
� N�0;
1n
Xj
ft(vtj )2
�
Letting v ti � N(0; �2t ). . .
�2t+1 =
1n
Xj
ft(vtj )2
= Efft(�t�)2g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 25 / 49
![Page 51: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/51.jpg)
(Wrong) Analysis
If i =2 S :
v t+1i =
1pn
Xj
Aij ft(vtj )
� N�0;
1n
Xj
ft(vtj )2
�
Letting v ti � N(0; �2t ). . .
�2t+1 =
1n
Xj
ft(vtj )2
= Efft(�t�)2g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 25 / 49
![Page 52: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/52.jpg)
(Wrong) Analysis
If i =2 S :
v t+1i =
1pn
Xj
Aij ft(vtj )
� N�0;
1n
Xj
ft(vtj )2
�
Letting v ti � N(0; �2t ). . .
�2t+1 =
1n
Xj
ft(vtj )2
= Efft(�t�)2g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 25 / 49
![Page 53: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/53.jpg)
(Wrong) Analysis
If i =2 S :
v t+1i =
1pn
Xj
Aij ft(vtj )
� N�0;
1n
Xj
ft(vtj )2
�
Letting v ti � N(0; �2t ). . .
�2t+1 =
1n
Xj
ft(vtj )2
= Efft(�t�)2g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 25 / 49
![Page 54: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/54.jpg)
(Wrong) AnalysisIf i 2 S :
v t+1i =
1pn
Xj2S
ft(vtj )
| {z }�t+1
+1pn
Xj =2S
Aij ft(vtj )
| {z }�N(0;�2
t+1)
where
�t+1 =1pn
Xj2S
ft(vtj )
=
�kpn
��1k
Xj2S
ft(vtj )
�= �Efft(�t + �t�)g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 26 / 49
![Page 55: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/55.jpg)
(Wrong) AnalysisIf i 2 S :
v t+1i =
1pn
Xj2S
ft(vtj )
| {z }�t+1
+1pn
Xj =2S
Aij ft(vtj )
| {z }�N(0;�2
t+1)
where
�t+1 =1pn
Xj2S
ft(vtj )
=
�kpn
��1k
Xj2S
ft(vtj )
�= �Efft(�t + �t�)g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 26 / 49
![Page 56: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/56.jpg)
(Wrong) AnalysisIf i 2 S :
v t+1i =
1pn
Xj2S
ft(vtj )
| {z }�t+1
+1pn
Xj =2S
Aij ft(vtj )
| {z }�N(0;�2
t+1)
where
�t+1 =1pn
Xj2S
ft(vtj )
=
�kpn
��1k
Xj2S
ft(vtj )
�= �Efft(�t + �t�)g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 26 / 49
![Page 57: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/57.jpg)
(Wrong) AnalysisIf i 2 S :
v t+1i =
1pn
Xj2S
ft(vtj )
| {z }�t+1
+1pn
Xj =2S
Aij ft(vtj )
| {z }�N(0;�2
t+1)
where
�t+1 =1pn
Xj2S
ft(vtj )
=
�kpn
��1k
Xj2S
ft(vtj )
�= �Efft(�t + �t�)g;� � N(0; 1)
Deshpande, Montanari Planted Cliques November 5, 2013 26 / 49
![Page 58: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/58.jpg)
Summarizing . . .
−4 −2 0 2 40
0.1
0.2
0.3
vti , i ∈ S
vti , i /∈ S
Deshpande, Montanari Planted Cliques November 5, 2013 27 / 49
![Page 59: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/59.jpg)
State Evolution
�t+1 = �E fft(�t + �t�)g�2t+1 = Efft(�t�)2g:
Using the optimal function ft(x) = e�tx��2t
�t+1 = �e�2t =2
�2t+1 = 1
Deshpande, Montanari Planted Cliques November 5, 2013 28 / 49
![Page 60: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/60.jpg)
State Evolution
�t+1 = �E fft(�t + �t�)g�2t+1 = Efft(�t�)2g:
Using the optimal function ft(x) = e�tx��2t
�t+1 = �e�2t =2
�2t+1 = 1
Deshpande, Montanari Planted Cliques November 5, 2013 28 / 49
![Page 61: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/61.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 62: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/62.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 63: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/63.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 64: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/64.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 65: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/65.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 66: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/66.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 67: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/67.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 68: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/68.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 69: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/69.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 70: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/70.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 71: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/71.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
If � < 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 72: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/72.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
If � < 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 73: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/73.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
If � < 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 74: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/74.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
If � < 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 75: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/75.jpg)
Fixed points develop below threshold!
If � > 1pe
µt+1
µt
If � < 1pe
µt+1
µt
Deshpande, Montanari Planted Cliques November 5, 2013 29 / 49
![Page 76: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/76.jpg)
Analysis is wrong but. . .
Theorem (Deshpande, Montanari, 2013)
If jS j = k � (1 + ")pn=e, there exists an O(n2 log n) time algorithm that
identifies S with high probability.
. . . so we modify the algorithm.
Deshpande, Montanari Planted Cliques November 5, 2013 30 / 49
![Page 77: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/77.jpg)
What algorithm?
Slight modification to iterative scheme:
(v ti )i2[n] ! (v ti!j)i ;j2[n]
v t+1i!j =
1pn
X6̀=i ;j
Ai` ft(vt`!i ):
Analysis is exact as n!1.
Deshpande, Montanari Planted Cliques November 5, 2013 31 / 49
![Page 78: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/78.jpg)
What algorithm?
Slight modification to iterative scheme:
(v ti )i2[n] ! (v ti!j)i ;j2[n]
v t+1i!j =
1pn
X6̀=i ;j
Ai` ft(vt`!i ):
Analysis is exact as n!1.
Deshpande, Montanari Planted Cliques November 5, 2013 31 / 49
![Page 79: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/79.jpg)
What algorithm?
Slight modification to iterative scheme:
(v ti )i2[n] ! (v ti!j)i ;j2[n]
v t+1i!j =
1pn
X6̀=i ;j
Ai` ft(vt`!i ):
Analysis is exact as n!1.
Deshpande, Montanari Planted Cliques November 5, 2013 31 / 49
![Page 80: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/80.jpg)
What algorithm?
Slight modification to iterative scheme:
(v ti )i2[n] ! (v ti!j)i ;j2[n]
v t+1i!j =
1pn
X6̀=i ;j
Ai` ft(vt`!i ):
Analysis is exact as n!1.
Deshpande, Montanari Planted Cliques November 5, 2013 31 / 49
![Page 81: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/81.jpg)
Fixing the heuristic
LemmaLet (ft(z))t�0 be a sequence of polynomials. Then, for every fixed t, andbounded, continuous function : R! R the following limit holds inprobability:
limn!1
1pn
Xi2S
(v ti!j) = �Ef (�t + �t�)g;
limn!1
1n
Xi2[n]nS
(v ti!j) = Ef (�t�)g;
where � � N(0; 1).
Deshpande, Montanari Planted Cliques November 5, 2013 32 / 49
![Page 82: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/82.jpg)
Proof Technique
Key ideas:
Expand v ti!j for polynomial ft(�)
Wrong analysis works if A! At
Deshpande, Montanari Planted Cliques November 5, 2013 33 / 49
![Page 83: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/83.jpg)
Proof Technique
Key ideas:
Expand v ti!j for polynomial ft(�)
Wrong analysis works if A! At
Deshpande, Montanari Planted Cliques November 5, 2013 33 / 49
![Page 84: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/84.jpg)
Proof Technique
Key ideas:
Expand v ti!j for polynomial ft(�)
Wrong analysis works if A! At
Deshpande, Montanari Planted Cliques November 5, 2013 33 / 49
![Page 85: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/85.jpg)
Proof Technique - Expanding v t
Let ft(x) = x2; v0i!j = 1
v1i!j =
Xk 6=j
Aik :i
k
j
Deshpande, Montanari Planted Cliques November 5, 2013 34 / 49
![Page 86: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/86.jpg)
Proof Technique - Expanding v t
v2i!j =
Xk 6=j
Aik(v1k!i )
2
=Xk 6=j
Aik
�X` 6=i
Ak`
��Xm 6=i
Akm
�=Xk 6=j
X` 6=i
Xm 6=i
AikAk`Akm:
i
j
k
` m
Deshpande, Montanari Planted Cliques November 5, 2013 35 / 49
![Page 87: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/87.jpg)
Proof Technique - Expanding v t
v2i!j =
Xk 6=j
Aik(v1k!i )
2
=Xk 6=j
Aik
�X` 6=i
Ak`
��Xm 6=i
Akm
�=Xk 6=j
X` 6=i
Xm 6=i
AikAk`Akm:
i
j
k
` m
Deshpande, Montanari Planted Cliques November 5, 2013 35 / 49
![Page 88: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/88.jpg)
Proof Technique - Expanding v t
v2i!j =
Xk 6=j
Aik(v1k!i )
2
=Xk 6=j
Aik
�X` 6=i
Ak`
��Xm 6=i
Akm
�=Xk 6=j
X` 6=i
Xm 6=i
AikAk`Akm:
i
j
k
` m
Deshpande, Montanari Planted Cliques November 5, 2013 35 / 49
![Page 89: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/89.jpg)
Proof Technique
v t+1i!j =
Xk 6=i
Aik ft(vtk!i ):
i
k
`m
o
p
qr
s w
�t+1i!j =
Xk 6=i
Atik ft(�
tk!i ):
i
k
`m
o
p
qr
s w
Deshpande, Montanari Planted Cliques November 5, 2013 36 / 49
![Page 90: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/90.jpg)
Proof Technique - a Combinatorial Lemma
Lemma
v ti!j =X
T2T ti!j
A(T )Γ(T )v0(T )
where T ti!j consists rooted, labeled trees that:
1 have maximum depth t.2 do not backtrack.
(Similarly for the �ti!j)
Deshpande, Montanari Planted Cliques November 5, 2013 37 / 49
![Page 91: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/91.jpg)
Proof Technique - Moment Method
v t+1i!j =
Xk 6=i
Aik ft(vtk!i ):
i
k
`m
o
p
qr
s w
�t+1i!j =
Xk 6=i
Atik ft(�
tk!i ):
i
k
`m
o
p
qr
s w
limn!1 Moments of v t+1 = Moments of �t+1.
Deshpande, Montanari Planted Cliques November 5, 2013 38 / 49
![Page 92: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/92.jpg)
Proof Technique - Moment Method
v t+1i!j =
Xk 6=i
Aik ft(vtk!i ):
i
k
`m
o
p
qr
s w
�t+1i!j =
Xk 6=i
Atik ft(�
tk!i ):
i
k
`m
o
p
qr
s w
limn!1 Moments of v t+1 = Moments of �t+1.
Deshpande, Montanari Planted Cliques November 5, 2013 38 / 49
![Page 93: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/93.jpg)
Progress(4)
Complexity
k
n2 logn
2 log2 n
n2
2√n log n
n2 log n
C√n1.261
√n
Spectral threshold =√n
√ne
Deshpande, Montanari Planted Cliques November 5, 2013 39 / 49
![Page 94: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/94.jpg)
Is this threshold fundamental?
Rest of the talk: perhaps
Deshpande, Montanari Planted Cliques November 5, 2013 40 / 49
![Page 95: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/95.jpg)
The “Hidden Set” Problem
Given Gn = ([n];En)
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S ;
Q0 otherwise.
Aij ∼ Q1
Aij ∼ Q0
Problem: Given edge labels (Aij)(i ;j)2En , identify S
Deshpande, Montanari Planted Cliques November 5, 2013 41 / 49
![Page 96: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/96.jpg)
The “Hidden Set” Problem
Given Gn = ([n];En)
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S ;
Q0 otherwise.
Aij ∼ Q1
Aij ∼ Q0
Problem: Given edge labels (Aij)(i ;j)2En , identify S
Deshpande, Montanari Planted Cliques November 5, 2013 41 / 49
![Page 97: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/97.jpg)
The “Hidden Set” Problem
Given Gn = ([n];En)
A Set ! S � [n]
Data ! Aij �(Q1 if i ; j 2 S ;
Q0 otherwise.
S
S
A =
Problem: Given edge labels (Aij)(i ;j)2En , identify S
Deshpande, Montanari Planted Cliques November 5, 2013 42 / 49
![Page 98: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/98.jpg)
“Local” Algorithms
A t-local algorithm computes:
Estimate at i :
bu(i) = F(ABall(i ;t))
Deshpande, Montanari Planted Cliques November 5, 2013 43 / 49
![Page 99: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/99.jpg)
The Sparse Graph Analogue
Gn = ([n];En); n � 1 satisfies:I locally tree-likeI regular degree ∆
FurtherI Q1 = �+1, Q0 = 1
2�+1 + 12��1
Aij ∼ Q1
Aij ∼ Q0
What can local algorithms achieve?
Deshpande, Montanari Planted Cliques November 5, 2013 44 / 49
![Page 100: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/100.jpg)
The Sparse Graph Analogue
Gn = ([n];En); n � 1 satisfies:I locally tree-likeI regular degree ∆
FurtherI Q1 = �+1, Q0 = 1
2�+1 + 12��1
Aij ∼ Q1
Aij ∼ Q0
What can local algorithms achieve?
Deshpande, Montanari Planted Cliques November 5, 2013 44 / 49
![Page 101: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/101.jpg)
The Sparse Graph Analogue
Gn = ([n];En); n � 1 satisfies:I locally tree-likeI regular degree ∆
FurtherI Q1 = �+1, Q0 = 1
2�+1 + 12��1
Aij ∼ Q1
Aij ∼ Q0
What can local algorithms achieve?
Deshpande, Montanari Planted Cliques November 5, 2013 44 / 49
![Page 102: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/102.jpg)
The Sparse Graph Analogue
Gn = ([n];En); n � 1 satisfies:I locally tree-likeI regular degree ∆
FurtherI Q1 = �+1, Q0 = 1
2�+1 + 12��1
Aij ∼ Q1
Aij ∼ Q0
What can local algorithms achieve?
Deshpande, Montanari Planted Cliques November 5, 2013 44 / 49
![Page 103: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/103.jpg)
What can we hope for?
If jS j = C np∆:
bSnaive = Random set of size jS j
) 1nEfbSnaive4Sg = Θ
�1p∆
�:
Deshpande, Montanari Planted Cliques November 5, 2013 45 / 49
![Page 104: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/104.jpg)
What can we hope for?
If jS j = C np∆:
Poisson bound ) for any local algorithm:
1nEfbS4Sg � e�C
0p
∆:
Deshpande, Montanari Planted Cliques November 5, 2013 46 / 49
![Page 105: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/105.jpg)
A result for local algorithms. . .
Theorem (Deshpande, Montanari, 2013)Let Gn converge locally to ∆�regular tree:If jS j � (1 + ") np
e∆there exists a local algorithm achieving
1nEfS4bSg � e�Θ(
p∆):
Conversely, if jS j � (1� ") npe∆
every local algorithm suffers
1nEfS4bSg � Θ
�1p∆
�
With ∆ = n � 1 we recover the complete graph result!
Deshpande, Montanari Planted Cliques November 5, 2013 47 / 49
![Page 106: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/106.jpg)
A result for local algorithms. . .
Theorem (Deshpande, Montanari, 2013)Let Gn converge locally to ∆�regular tree:If jS j � (1 + ") np
e∆there exists a local algorithm achieving
1nEfS4bSg � e�Θ(
p∆):
Conversely, if jS j � (1� ") npe∆
every local algorithm suffers
1nEfS4bSg � Θ
�1p∆
�
With ∆ = n � 1 we recover the complete graph result!
Deshpande, Montanari Planted Cliques November 5, 2013 47 / 49
![Page 107: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/107.jpg)
A result for local algorithms. . .
Theorem (Deshpande, Montanari, 2013)Let Gn converge locally to ∆�regular tree:If jS j � (1 + ") np
e∆there exists a local algorithm achieving
1nEfS4bSg � e�Θ(
p∆):
Conversely, if jS j � (1� ") npe∆
every local algorithm suffers
1nEfS4bSg � Θ
�1p∆
�
With ∆ = n � 1 we recover the complete graph result!
Deshpande, Montanari Planted Cliques November 5, 2013 47 / 49
![Page 108: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/108.jpg)
To conclude. . .
I Message-passing algorithm performs “weighted” counts ofnon-reversing trees
I Such structures have been used elsewhere:� Clustering sparse networks: [Krzakala et al. 2013]� Compressed sensing: [Bayati, Lelarge, Montanari 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 48 / 49
![Page 109: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/109.jpg)
To conclude. . .
I Message-passing algorithm performs “weighted” counts ofnon-reversing trees
I Such structures have been used elsewhere:� Clustering sparse networks: [Krzakala et al. 2013]� Compressed sensing: [Bayati, Lelarge, Montanari 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 48 / 49
![Page 110: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/110.jpg)
To conclude. . .
I Message-passing algorithm performs “weighted” counts ofnon-reversing trees
I Such structures have been used elsewhere:� Clustering sparse networks: [Krzakala et al. 2013]� Compressed sensing: [Bayati, Lelarge, Montanari 2013]
Deshpande, Montanari Planted Cliques November 5, 2013 48 / 49
![Page 111: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/111.jpg)
To conclude. . .
I What is the “dense” analogue for local algorithms?
I What about other structural properties?
Thank you!
Deshpande, Montanari Planted Cliques November 5, 2013 49 / 49
![Page 112: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/112.jpg)
To conclude. . .
I What is the “dense” analogue for local algorithms?
I What about other structural properties?
Thank you!
Deshpande, Montanari Planted Cliques November 5, 2013 49 / 49
![Page 113: Planted Cliques, Iterative Thresholding and Message ...yash/SLIDES/simons.pdf · Deshpande, Montanari Planted Cliques November 5, 2013 21 / 49 Seemsmuchharderthanitlooks! I “Statisticalalgorithms”](https://reader030.vdocument.in/reader030/viewer/2022040508/5e4d936a445f5c11c1732b1e/html5/thumbnails/113.jpg)
To conclude. . .
I What is the “dense” analogue for local algorithms?
I What about other structural properties?
Thank you!
Deshpande, Montanari Planted Cliques November 5, 2013 49 / 49