Filtering and Sampling Graph Signals, and its Application to Compressive Spectral Clustering
|
|
- Gavin Brown
- 5 years ago
- Views:
Transcription
1 Filtering and Sampling Graph Signals, and its Application to Compressive Spectral Clustering Nicolas Tremblay (1,2), Gilles Puy (1), Rémi Gribonval (1), Pierre Vandergheynst (1,2) (1) PANAMA Team, INRIA Rennes, France (2) Signal Processing Laboratory 2, EPFL, Switzerland
2 Introduction to GSP Graph sampling Application to clustering Conclusion Why graph signal processing? N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
3 Introduction to GSP Graph Fourier Transform Graph filtering Graph sampling Application to clustering What is Spectral Clustering? Compressive Spectral Clustering A toy experiment Experiments on the SBM Conclusion N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
4 Introduction to graph signal processing : graph Fourier transform N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
5 What s a graph signal? N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
6 Three useful matrices The adjacency matrix : The degree matrix : W = S = The Laplacian matrix : L = S W = N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
7 Three useful matrices The adjacency matrix : The degree matrix : W = S = The Laplacian matrix : L = S W = N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
8 What s a graph Fourier transform? [Hammond 11] L = S W = UΛU U is the Fourier basis of the graph the Fourier transform of a signal x reads : ˆx = U x Λ = Diag(λ 1, λ 2,, λ N ) the spectrum A low frequency Fourier mode A high frequency Fourier mode N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
9 The graph Fourier transform encodes the structure of the graph Slide courtesy of D. Shuman N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
10 Introduction to graph signal processing : filtering graph signals N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
11 Graph filtering 1 Given a filter function h defined in the Fourier space. g(λ) h λ In the node space, the signal x filtered by h reads : x h = U h(λ) U x = Hx N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
12 Graph filtering 1 Given a filter function h defined in the Fourier space. g(λ) h λ In the node space, the signal x filtered by h reads : Problem : this costs L s diagonalisation [O(N 3 )]. x h = U h(λ) U x = Hx N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
13 Graph filtering 1 Given a filter function h defined in the Fourier space. g(λ) h λ In the node space, the signal x filtered by h reads : Problem : this costs L s diagonalisation [O(N 3 )]. x h = U h(λ) U x = Hx Solution : we use a poly approx of order p of h : p h(λ) = α l λ l h(λ). Indeed, in this case : p Hx = U h(λ)u x = U α l Λ l U x = l=1 l=1 l=1 p α l L l x Hx Only involves matrix-vector multiplications [costs O(pN)]. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
14 A few applications Tikhonov regularization for denoising : argmin f { f y γf Lf } Wavelet denoising : argmin a { f W a γ a 1,µ } Compression via filterbanks, etc. Slide courtesy of D. Shuman N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
15 Introduction to GSP Graph Fourier Transform Graph filtering Graph sampling Application to clustering What is Spectral Clustering? Compressive Spectral Clustering A toy experiment Experiments on the SBM Conclusion N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
16 Sampling a graph signal consists in : 1. choosing a subset of nodes 2. measuring the signal on these nodes only N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
17 Sampling a graph signal consists in : 1. choosing a subset of nodes 2. measuring the signal on these nodes only N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
18 Sampling a graph signal consists in : 1. choosing a subset of nodes 2. measuring the signal on these nodes only How to reconstruct the original signal? Basically, we need : 1. a (low-dimensional) model for the signal to sample 2. a method to choose the nodes to sample 3. a decoder that exactly recovers the signal given its samples N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
19 Smoothness assumption In 1D signal processing, a smooth signal has most of its energy at low frequencies Smooth signal in time Fourier transform N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
20 Smoothness assumption In 1D signal processing, a smooth signal has most of its energy at low frequencies Smooth signal in time Fourier transform Definition (Bandlimited graph signal [Puy 15, Chen 15, Anis 16, Segarra 15] ) A k-bandlimited signal x R N on G is a signal that satisfies, for some ˆα R k x = U k ˆα, N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
21 Sampling band-limited graph signals Preparation : Associate to each node i a probability p i to draw this node. This defines a probability distribution p R N. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
22 Sampling band-limited graph signals Preparation : Associate to each node i a probability p i to draw this node. This defines a probability distribution p R N. Sampling procedure : draw n nodes according to p : {ω i } i [1,n]. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
23 Sampling band-limited graph signals Preparation : Associate to each node i a probability p i to draw this node. This defines a probability distribution p R N. Sampling procedure : draw n nodes according to p : {ω i } i [1,n]. We create a matrix M that measures the signal x only on the selected nodes : { 1 if j = ωi M ij := 0 otherwise, For any signal x R N on G, its sampled version is y = Mx (it has size n < N). N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
24 Optimizing the sampling distribution Some nodes are more important to sample than others. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
25 Optimizing the sampling distribution Some nodes are more important to sample than others. For any signal x, remember that U k x is the energy of x on the first k 2 frequencies. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
26 Optimizing the sampling distribution Some nodes are more important to sample than others. For any signal x, remember that U k x is the energy of x on the first k 2 frequencies. Then : 1. For each node i, construct the Dirac δ i centered at node i. 2. Compute U k δ i 2 (we have 0 U k δ i 2 1). N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
27 Optimizing the sampling distribution Some nodes are more important to sample than others. For any signal x, remember that U k x is the energy of x on the first k 2 frequencies. Then : 1. For each node i, construct the Dirac δ i centered at node i. 2. Compute U k δ i 2 (we have 0 U k δ i 2 1). If U k δ i 2 1 : there exists a smooth signal concentrated on node i. Node i is important. If U k δ i 2 0 : no smooth signal has energy concentrated on node i. Node i can be sampled with less probability. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
28 The graph weighted coherence We measure the quality of p with the graph weighted coherence. Definition (Graph weighted coherence) Let p R n represent a sampling distribution on {1,..., N}. The graph weighted coherence of order k for the pair (G, p) is { } νp k := max p 1/2 i U k δ i 1 i N 2. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
29 How many nodes to select? Theorem (Restricted isometry property) Let M be a random subsampling matrix constructed using the sampling distribution p. For any δ, ɛ (0, 1), with probability at least 1 ɛ, (1 δ) x 1 x MP 1/2 (x 1 x 2) 2 (1 + δ) x 1 x n 2 for all x 1, x 2 span(u k ) provided that n 3 δ 2 (νk p ) 2 log ( ) 2k. ɛ N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
30 How many nodes to select? Theorem (Restricted isometry property) Let M be a random subsampling matrix constructed using the sampling distribution p. For any δ, ɛ (0, 1), with probability at least 1 ɛ, (1 δ) x 1 x MP 1/2 (x 1 x 2) 2 (1 + δ) x 1 x n 2 for all x 1, x 2 span(u k ) provided that n 3 δ 2 (νk p ) 2 log ( ) 2k. ɛ Let s minimize ν k p! Its lower bound, k, may always be reached for p : i [1, N] p i = U k δ i 2 2 /k N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
31 How many nodes to select? Theorem (Restricted isometry property) Let M be a random subsampling matrix constructed using the sampling distribution p. For any δ, ɛ (0, 1), with probability at least 1 ɛ, (1 δ) x 1 x MP 1/2 (x 1 x 2) 2 (1 + δ) x 1 x n 2 for all x 1, x 2 span(u k ) provided that n 3 δ 2 (νk p ) 2 log ( ) 2k. ɛ Let s minimize ν k p! Its lower bound, k, may always be reached for p : i [1, N] p i = U k δ i 2 2 /k With p, one needs n k log (k) up to the log factor, it is optimal! N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
32 How many nodes to select? Theorem (Restricted isometry property) Let M be a random subsampling matrix constructed using the sampling distribution p. For any δ, ɛ (0, 1), with probability at least 1 ɛ, (1 δ) x 1 x MP 1/2 (x 1 x 2) 2 (1 + δ) x 1 x n 2 for all x 1, x 2 span(u k ) provided that n 3 δ 2 (νk p ) 2 log ( ) 2k. ɛ Let s minimize ν k p! Its lower bound, k, may always be reached for p : i [1, N] p i = U k δ i 2 2 /k With p, one needs n k log (k) up to the log factor, it is optimal! We have an efficient algorithm that estimates p in O(pN log N)! N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
33 Reconstruction We sampled the signal x R N, i.e., we measured y = Mx + n (n R n models noise). The goal is to estimate x from y. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
34 Reconstruction We sampled the signal x R N, i.e., we measured y = Mx + n (n R n models noise). The goal is to estimate x from y. We propose to solve (links with the SSL literature [Chapelle 10, Fu 12]) P 1/2 min z R N Ω (Mz y) 2 + γ z g(l)z, 2 where γ > 0 and g : R R is a nonnegative and nondecreasing poly function. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
35 Reconstruction Solving P 1/2 min z R N Ω (Mz y) 2 + γ z g(l)z, 2 can be done, e.g., by gradient descent or conjugate gradient. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
36 Reconstruction Solving P 1/2 min z R N Ω (Mz y) 2 + γ z g(l)z, 2 can be done, e.g., by gradient descent or conjugate gradient. It is fast as it involves only matrix-vector multiplications with sparse matrices. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
37 Reconstruction Solving P 1/2 min z R N Ω (Mz y) 2 + γ z g(l)z, 2 can be done, e.g., by gradient descent or conjugate gradient. It is fast as it involves only matrix-vector multiplications with sparse matrices. We proved that the result is accurate and stable to noise : The quality of the reconstruction depends on the eigengap ratio g(λ k )/g(λ k+1 ). γ should be adjusted with the signal-to-noise ratio. In absence of noise, the reconstruction quality improves when g(λ k )/g(λ k+1 ) 0 and γ 0. If g(λ k ) = 0 and g(λ k+1 ) > 0, we have exact recovery. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
38 Recap Given a graph and its Laplacian matrix L. Given any graph signal x defined on this graph, one can : 1. filter this signal with any filter h(λ) : x h = U h(λ) U [O(N 3 )], 2. fast filter it w/ poly approx h(λ) p l=1 α lλ l : x h p l=1 α ll l x [O(pN)]. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
39 Recap Given a graph and its Laplacian matrix L. Given any graph signal x defined on this graph, one can : 1. filter this signal with any filter h(λ) : x h = U h(λ) U [O(N 3 )], 2. fast filter it w/ poly approx h(λ) p l=1 α lλ l : x h p l=1 α ll l x [O(pN)]. Given a k-bandlimited graph signal x defined on this graph, one can : 1. estimate the optimal probability distrib pi = U k δ i 2 /k [O(pN log N)] 2 2. sample n = O(k log k) nodes from this distribution 3. measure the signal y = Mx R n 4. reconstruct the signal : [O(pN)] x rec = argmin P 1/2 Ω (Mz y) 2 + γ z g(l)z z R N 2 N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
40 Introduction to GSP Graph Fourier Transform Graph filtering Graph sampling Application to clustering What is Spectral Clustering? Compressive Spectral Clustering A toy experiment Experiments on the SBM Conclusion N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
41 Application to clustering : What is Spectral Clustering? N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
42 Given a series of N objects : N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
43 Given a series of N objects : 1/ Find adapted descriptors N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
44 Given a series of N objects : 1/ Find adapted descriptors 2/ Cluster N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
45 From the N objects, one creates : N vectors : x 1, x 2,, x N and their distance matrix R N N. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
46 From the N objects, one creates : N vectors : x 1, x 2,, x N and their distance matrix R N N. Goal of clustering : assign a label c(i) = 1,, k to each object i in order to organize / simplify / analyze the data. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
47 From the N objects, one creates : N vectors : x 1, x 2,, x N and their distance matrix R N N. Goal of clustering : assign a label c(i) = 1,, k to each object i in order to organize / simplify / analyze the data. There exists two different general types of methods : methods directly based on the x i and/or like k-means or hierarchical clustering. graph-based methods. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
48 Graph construction from the distance matrix Create a graph G = (V, E) : each node in V is one of the N objects each pair of nodes (i, j) is connected if (i, j) is small enough. For example, two connectivity possibilities : Gaussian kernel : 1. all pairs of nodes are connected with links of weights exp( (i, j)/σ) 2. remove all links of weight inferior to ɛ k nearest neighbors : connect each node to its k nearest neighbors. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
49 The clustering problem now states : Given the graph G representing the similarity between the N objects, find a partition of all nodes into k clusters. Many methods exist [Fortunato 10] : Modularity (or other cost-function) optimisation methods [Newman 06] Random walk methods [Schaub 12] Methods inspired from statistical physics [Krzakala 13], information theory [Rosvall 08]... spectral methods... N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
50 The classical spectral clustering (SC) algorithm [Von Luxburg 06] : Given the N-node graph G of laplacian matrix L : 1. Compute L s first k eigenvectors : U k = (u 1 u 2 u k ). 2. Consider each node i as a point in R k : f i = U k δ i. 3. Run k-means with the Euclidean distance : D ij = f i f j and obtain k clusters. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
51 The classical spectral clustering (SC) algorithm [Von Luxburg 06] : Given the N-node graph G of laplacian matrix L : 1. Compute L s first k eigenvectors : U k = (u 1 u 2 u k ). 2. Consider each node i as a point in R k : f i = U k δ i. 3. Run k-means with the Euclidean distance : D ij = f i f j and obtain k clusters. Definition : Let us call D ij the spectral clustering distance. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
52 What s the point of using a graph? N points in d = 2 dims. Result with k-means (k=2) on : After creating a graph, partial diagonalisation of L and running k-means (k=2) on D : N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
53 Application to clustering : Compressive Spectral Clustering N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
54 Our goal Problem : N and/or k large, two main bottlenecks : 1. partial eigendecomposition of (sparse) Laplacian (e.g. restarted Arnoldi) [at least O(k 3 + Nk 2 )] [Chen 11a] 2. high-dimensional k-means [O(Nk 2 )]. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
55 Our goal Problem : N and/or k large, two main bottlenecks : 1. partial eigendecomposition of (sparse) Laplacian (e.g. restarted Arnoldi) [at least O(k 3 + Nk 2 )] [Chen 11a] 2. high-dimensional k-means [O(Nk 2 )]. Goal : SC in high dimensions : with N 10 6 nodes and/or k 100. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
56 Our goal Problem : N and/or k large, two main bottlenecks : 1. partial eigendecomposition of (sparse) Laplacian (e.g. restarted Arnoldi) [at least O(k 3 + Nk 2 )] [Chen 11a] 2. high-dimensional k-means [O(Nk 2 )]. Goal : SC in high dimensions : with N 10 6 nodes and/or k 100. Contribution : an algorithm that approximates the true SC solution with controlled relative error with a running time in O(k 2 log 2 k + pn(log(n) + k)). N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
57 Main ideas of Compressive Spectral Clustering (CSC) : CSC is based on two main observations : 1. SC does not need explicitly f i = U k δ i, but only D ij = f i f j 2. each cluster indicator function c j R N is in fact approx. k-bandlimited! j [1, k] c j is close to span(u k ) CSC follows 4 steps : 1. Estimate D ij by filtering d random graph signals, 2. Sample n nodes out of the N available ones, 3. Run low-dim k-means on these n nodes to obtain c r j R n, 4. Reconstruct each reduced cluster indicator function c r j back on the whole graph to obtain c j, as desired. (Steps 2 to 4 already covered!) Step 1 : How to estimate D ij without computing U k? N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
58 Remember : the classical spectral clustering algorithm Given the N-node graph G of Laplacian matrix L : 1. Compute L s first k eigenvectors : U k = (u 1 u 2 u k ). 2. Consider each node i as a point in R k : f i = U k δ i. 3. Run k-means with D ij = f i f j and obtain k clusters. Our goal : Estimate D ij without computing exactly U k. D ij = U k (δ i δ j ) = U 1.5 k δ ij 1 = U k U k δ ij 0.5 = U h λk (Λ) U 0 δ ij h λk (λ) λ k 1 2 = H λk δ ij. λ N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
59 Fast filtering [Hammond, ACHA 11] In practice, we use a poly approx of order p of h λk : h λk = p α l λ l h λk. l=1 h λk (λ) 1.5 ideal 1 m=100 p m=20 p 0.5 m=5 p λ k 1 2 λ Such that : D ij = H λk δ ij = lim p H λk δ ij N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
60 Norm conservation result [Tremblay 16a, Ramasamy 15 ] The spectral distance reads : D ij = H λk δ ij = lim p Hλk δ ij N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
61 Norm conservation result [Tremblay 16a, Ramasamy 15 ] The spectral distance reads : D ij = H λk δ ij = lim p Hλk δ ij Let R = (r 1 r 2 r d ) R N d be a random Gaussian matrix, i.e. a collection of d random graph signals, with 0 mean and var. 1/d. We define f i = ( H λk R) δ i R d and D ij = f i f j N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
62 Norm conservation result [Tremblay 16a, Ramasamy 15 ] The spectral distance reads : D ij = H λk δ ij = lim p Hλk δ ij Let R = (r 1 r 2 r d ) R N d be a random Gaussian matrix, i.e. a collection of d random graph signals, with 0 mean and var. 1/d. We define f i = ( H λk R) δ i R d and D ij = f i f j Theorem (Norm conservation theorem in the case of infinite p) Let ɛ > 0, if d > d 0 log N/ɛ 2, then, with proba > 1 1/N, we have : (i, j) [1, N] 2 (1 ɛ)d ij D ij (1 + ɛ)d ij. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
63 Norm conservation result [Tremblay 16a, Ramasamy 15 ] The spectral distance reads : D ij = H λk δ ij = lim p Hλk δ ij Let R = (r 1 r 2 r d ) R N d be a random Gaussian matrix, i.e. a collection of d random graph signals, with 0 mean and var. 1/d. We define f i = ( H λk R) δ i R d and D ij = f i f j Theorem (Norm conservation theorem in the case of infinite p) Let ɛ > 0, if d > d 0 log N/ɛ 2, then, with proba > 1 1/N, we have : (i, j) [1, N] 2 (1 ɛ)d ij D ij (1 + ɛ)d ij. Consequence : to estimate D ij with no partial diagonalisation of L, fast filter only d log N random signals! N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
64 How to quickly estimate λ k, the sole unknown of the fast filtering operation? Goal : given a SDP L, estimate its k-th eigenvalue as fast as possible. We use eigencount techniques [Napoli 13] (also based on polynomial filtering of random vectors!) : given the interval [0, b], get an approximation of the number of enclosed eigenvalues. And find λ k by dichotomy on b. done in [O(pN log N)] N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
65 The CSC algorithm [Tremblay 16b, Puy 16 ] 1. Estimate λ k, the k-th eigenvalue of L. 2. Generate d random graph signals in matrix R R N d. 3. Filter them with H λk and treat each node i as a point in R d : f i = δ i H λk R. If d log N, we prove that D ij = f i f j D ij. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
66 The CSC algorithm [Tremblay 16b, Puy 16 ] 1. Estimate λ k, the k-th eigenvalue of L. 2. Generate d random graph signals in matrix R R N d. 3. Filter them with H λk and treat each node i as a point in R d : f i = δ i H λk R. If d log N, we prove that D ij = f i f j D ij. Next steps (sampling) : 4. sample n nodes from p 5. run k-means on the n associated feature vectors and obtain {cj r } j=1:k 6. reconstruct all k indicator functions {c j } j=1:k If n k log k and c r j = Mc j, we prove that we control the reconstruction error. N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
67 Application to clustering : A toy experiment N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
68 Introduction to GSP Graph sampling Application to clustering Conclusion SC on a toy example N = 1000, k = 2 Com 1 : 300 nodes } } Com 2 : 700 nodes Compute U2 = (u1, u2 ) N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
69 Introduction to GSP Graph sampling Application to clustering Conclusion SC on a toy example 2 fi = U> 2 δi R : N = 1000, k = 2 Com 1 : 300 nodes } } Com 2 : 700 nodes Compute U2 = (u1, u2 ) N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
70 Introduction to GSP Graph sampling Application to clustering Conclusion SC on a toy example 2 fi = U> 2 δi R : N = 1000, k = 2 Com 1 : 300 nodes } } Com 2 : 700 nodes Dij : Compute U2 = (u1, u2 ) N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
71 Introduction to GSP Graph sampling Application to clustering Conclusion SC on a toy example 2 fi = U> 2 δi R : N = 1000, k = 2 Com 1 : 300 nodes 0.5 } k-means 0 perf = } Com 2 : 700 nodes Dij : Compute U2 = (u1, u2 ) N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
72 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example 1. estimate λ2 and p 2. gen. d = 3 random graph signals N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
73 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
74 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i D ij ' Dij : N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
75 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i sample n = 3 nodes from p D ij ' Dij : N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
76 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i 4. sample n = 3 nodes from p D ij ' Dij : N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
77 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : k-means estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i sample n = 3 nodes from p D ij ' Dij : 5. run low-dim. k-means N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
78 Introduction to GSP Graph sampling Application to clustering Conclusion CSC on the same toy example f i R3 : -0.4 after interpolation estimate λ2 and p 2. gen. d = 3 random graph signals 3. low-pass filter them : f i sample n = 3 nodes from p perf = D ij ' Dij : 5. run low-dim. k-means 6. reconstruct the result N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
79 Application to clustering : Experiments on the SBM N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
80 Experiments The Stochastic Block Model (SBM) : C 1 C 2 C k N nodes and k communities of equal size N/k. { { q 1 q 2 q 2 q 1 q 2 q 2 { q 2 q 2 q 1 proba q 1 if in same community proba q 2 if not. define the ratio ɛ = q 2/q 1 SBM fully defined by ɛ and average degree s. define critical ratio ɛ c = (s s)/(s + s(k 1)) [Decelle 11] N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
81 Experiments The Stochastic Block Model (SBM) : C 1 C 2 C k N nodes and k communities of equal size N/k. { { q 1 q 2 q 2 q 1 q 2 q 2 { q 2 q 2 q 1 proba q 1 if in same community proba q 2 if not. define the ratio ɛ = q 2/q 1 SBM fully defined by ɛ and average degree s. define critical ratio ɛ c = (s s)/(s + s(k 1)) [Decelle 11] Experiments with N = 10 3, k = 20, s = 16, wrt to different parameters : Recovery performance SC n = k log(k) n = 2 k log(k) n = 3 k log(k) n = 4 k log(k) ǫ c 0.15 ǫ Recovery performance 1 SC 0.5 d = 2 log(n) d = 3 log(n) d = 4 log(n) d = 5 log(n) ǫ c 0.15 ǫ Recovery performance SC p = 10 p = 20 p = 50 p = ǫ c 0.15 ǫ N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
82 Experiments With params d = 4 log (k), n = 2k log (k), p = 50, γ = 10 3, and ɛ = ɛ c/4 : Recovery performance # of classes k N=10 4, CSC N=10 4, PM N=10 4, SC N=10 5, CSC N=10 5, PM N=10 5, SC N=10 6, CSC N=10 6, PM N=10 6, SC PM = Power Method [Lin 10, Boutsidis 15] Computation time (s) # of classes k N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
83 Experiments With params d = 4 log (k), n = 2k log (k), p = 50, γ = 10 3, and ɛ = ɛ c/4 : Recovery performance # of classes k N=10 4, CSC N=10 4, PM N=10 4, SC N=10 5, CSC N=10 5, PM N=10 5, SC N=10 6, CSC N=10 6, PM N=10 6, SC PM = Power Method [Lin 10, Boutsidis 15] Computation time (s) # of classes k On a real-world graph : Amazon graph with nodes and edges : SC CSC k=250 7h17m, h20m, 0.83 k=500 15h29m, h34m, h36m (eigs) k= at least 21 h 10h18m, 0.84 for k-means, unknown N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
84 Conclusion N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
85 Two main ideas Low-pass graph fast filtering of random signals : a way to by-pass the Laplacian s diagonalisation for learning tasks. Cluster indicator functions live in a low-dimensional space (are k-bandlimited) : we can use sampling schemes to recover them efficiently. Details of this work are found in : (Sampling part) Random sampling of bandlimited signals on graphs, ACHA A MATLAB toolbox is available at : grsamplingbox.gforge.inria.fr (Clustering part) Compressive Spectral Clustering, ICML A MATLAB toolbox is available at : cscbox.gforge.inria.fr N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
86 Links with literature Low-rank approximation : Nystrom methods [Sun 15], leverage scores [Mahoney 11] Machine Learning : semi-supervised learning [Chapelle 10], active learning [Fu 12, Gadde 14], coresets [Har-Peled 04, Frahling 08] Compressed sensing : variable density sampling [Puy 11] Other fast approximate SC algorithms : [Lin 10, Fowlkes 04, Wang 09, Chen 11a, Chen 11b] N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
87 Perspectives and difficult questions Two difficult questions (among others) : 1. Given a SDP matrix, how to estimate as fast as possible its k-th eigenvalue, and only that one? 2. How to choose automatically the appropriate polynomial order p? N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
88 Perspectives and difficult questions Two difficult questions (among others) : 1. Given a SDP matrix, how to estimate as fast as possible its k-th eigenvalue, and only that one? 2. How to choose automatically the appropriate polynomial order p? Perspectives 1. Rational filters instead of polynomial filters? [Shi 15, Isufi 16] 2. Smoother filters for better approximation? [Sakiyama 16] 3. How about if nodes are added one by one? 4. SBMO! [cf. E. Kaufmann] 5. Experiments shown were done with L = I D 1/2 WD 1/2. Test for L = D 1 2 ˆα D ˆα WD ˆα! [cf. R. Couillet] N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
89 [Ramasamy 15] Compressive spectral embedding : sidestepping... NIPS. [Fortunato 10] Community detection in graphs, Physics Reports [Newman 06] Modularity and community structure in networks, PNAS [Schaub 12] Markov dynamics as a zooming lens for multiscale..., Plos One [Krzakala 13] Spectral redemption : clustering sparse networks, PNAS [Rosvall 08] Maps of random walks on complex networks reveal..., PLOS One [Von Luxburg 06] A tutorial on spectral clustering, Statistics and Computing. [Chen 11a] Parallel spectral clustering in distributed systems, IEEE TPAMI [Lin 10] Power iteration clustering, ICML [Boutsidis 15] Spectral clustering via the power method - provably, ICML [Fowlkes 04] Spectral grouping using the nystrom method, IEEE TPAMI [Wang 09] Approximate spectral clustering, AKDDM [Chen 11b] Large scale spectral clustering with landmark-based..., CAI [Shuman 13] The emerging field of signal processing on graphs..., SPMag [Hammond 11] Wavelets on graphs via spectral graph theory, ACHA [Napoli 13] Efficient estimation of eigenvalue counts in an interval, arxiv [Tremblay 16a] Accelerated spectral clustering using graph..., ICASSP [Tremblay 16b] Compressive spectral clustering, ICML [Puy 16] Random sampling of bandlimited signals..., ACHA [Shi 15] Infinite impulse response graph filters in wireless sensor networks, SPL [Chen 15] Discrete Signal Processing on Graphs : Sampling Theory, TSP [Anis 16] Efficient Sampling Set Selection for Bandlimited Graph..., TSP [Segarra 15] Sampling of graph signals with successive local aggregations, TSP [Chapelle 10] Semi-Supervised Learning, The MIT Press [Fu 12] A survey on instance selection for active learning, KIS [Mahoney 11] Randomized algorithms for matrices and data, Found. and Trends in ML [Sun 15] A review of Nyström methods for large-scale machine learning, Inf. Fus. [Puy 11] On variable density compressive sampling, SPL [Gadde 14] Active Semi-supervised Learning Using Sampling Theory..., SIGKDD [Isufi 16] Distributed Time-Varying Graph Filtering, ArXiv [Sakiyama 16] Sp. Gr. Wav. and Filter Banks with Low Approximation Error, not published yet N. Tremblay Compressive Spectral Clustering Gdr ISIS, 17th of June / 47
Random Sampling of Bandlimited Signals on Graphs
Random Sampling of Bandlimited Signals on Graphs Pierre Vandergheynst École Polytechnique Fédérale de Lausanne (EPFL) School of Engineering & School of Computer and Communication Sciences Joint work with
More informationSampling, Inference and Clustering for Data on Graphs
Sampling, Inference and Clustering for Data on Graphs Pierre Vandergheynst École Polytechnique Fédérale de Lausanne (EPFL) School of Engineering & School of Computer and Communication Sciences Joint work
More informationRandom sampling of bandlimited signals on graphs
Random sampling of bandlimited signals on graphs Gilles Puy, Nicolas Tremblay, Rémi Gribonval, Pierre Vandergheynst To cite this version: Gilles Puy, Nicolas Tremblay, Rémi Gribonval, Pierre Vandergheynst
More informationarxiv: v1 [cs.ds] 5 Feb 2016
Compressive spectral clustering Nicolas Tremblay, Gilles Puy, Rémi Gribonval, and Pierre Vandergheynst arxiv:6.8v [cs.ds] 5 Feb 6 Abstract. Spectral clustering has become a popular technique due to its
More informationRandom sampling of bandlimited signals on graphs
Random sampling of bandlimited signals on graphs Gilles Puy, Nicolas Tremblay, Rémi Gribonval, Pierre Vandergheynst To cite this version: Gilles Puy, Nicolas Tremblay, Rémi Gribonval, Pierre Vandergheynst
More informationGraph sampling with determinantal processes
Graph sampling with determinantal processes Nicolas Tremblay, Pierre-Olivier Amblard, Simon Barthelme To cite this version: Nicolas Tremblay, Pierre-Olivier Amblard, Simon Barthelme. Graph sampling with
More informationarxiv: v3 [stat.ml] 29 May 2018
On Consistency of Compressive Spectral Clustering On Consistency of Compressive Spectral Clustering arxiv:1702.03522v3 [stat.ml] 29 May 2018 Muni Sreenivas Pydi Department of Electrical and Computer Engineering
More informationMLCC Clustering. Lorenzo Rosasco UNIGE-MIT-IIT
MLCC 2018 - Clustering Lorenzo Rosasco UNIGE-MIT-IIT About this class We will consider an unsupervised setting, and in particular the problem of clustering unlabeled data into coherent groups. MLCC 2018
More informationSpectral Clustering. Guokun Lai 2016/10
Spectral Clustering Guokun Lai 2016/10 1 / 37 Organization Graph Cut Fundamental Limitations of Spectral Clustering Ng 2002 paper (if we have time) 2 / 37 Notation We define a undirected weighted graph
More informationA PROBABILISTIC INTERPRETATION OF SAMPLING THEORY OF GRAPH SIGNALS. Akshay Gadde and Antonio Ortega
A PROBABILISTIC INTERPRETATION OF SAMPLING THEORY OF GRAPH SIGNALS Akshay Gadde and Antonio Ortega Department of Electrical Engineering University of Southern California, Los Angeles Email: agadde@usc.edu,
More informationarxiv: v1 [cs.it] 26 Sep 2018
SAPLING THEORY FOR GRAPH SIGNALS ON PRODUCT GRAPHS Rohan A. Varma, Carnegie ellon University rohanv@andrew.cmu.edu Jelena Kovačević, NYU Tandon School of Engineering jelenak@nyu.edu arxiv:809.009v [cs.it]
More informationSemi-Supervised Learning in Gigantic Image Collections. Rob Fergus (New York University) Yair Weiss (Hebrew University) Antonio Torralba (MIT)
Semi-Supervised Learning in Gigantic Image Collections Rob Fergus (New York University) Yair Weiss (Hebrew University) Antonio Torralba (MIT) Gigantic Image Collections What does the world look like? High
More informationSketching for Large-Scale Learning of Mixture Models
Sketching for Large-Scale Learning of Mixture Models Nicolas Keriven Université Rennes 1, Inria Rennes Bretagne-atlantique Adv. Rémi Gribonval Outline Introduction Practical Approach Results Theoretical
More informationSpectral Graph Wavelets on the Cortical Connectome and Regularization of the EEG Inverse Problem
Spectral Graph Wavelets on the Cortical Connectome and Regularization of the EEG Inverse Problem David K Hammond University of Oregon / NeuroInformatics Center International Conference on Industrial and
More informationData Analysis and Manifold Learning Lecture 7: Spectral Clustering
Data Analysis and Manifold Learning Lecture 7: Spectral Clustering Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline of Lecture 7 What is spectral
More informationDigraph Fourier Transform via Spectral Dispersion Minimization
Digraph Fourier Transform via Spectral Dispersion Minimization Gonzalo Mateos Dept. of Electrical and Computer Engineering University of Rochester gmateosb@ece.rochester.edu http://www.ece.rochester.edu/~gmateosb/
More informationLearning on Graphs and Manifolds. CMPSCI 689 Sridhar Mahadevan U.Mass Amherst
Learning on Graphs and Manifolds CMPSCI 689 Sridhar Mahadevan U.Mass Amherst Outline Manifold learning is a relatively new area of machine learning (2000-now). Main idea Model the underlying geometry of
More informationSpectral Clustering. Spectral Clustering? Two Moons Data. Spectral Clustering Algorithm: Bipartioning. Spectral methods
Spectral Clustering Seungjin Choi Department of Computer Science POSTECH, Korea seungjin@postech.ac.kr 1 Spectral methods Spectral Clustering? Methods using eigenvectors of some matrices Involve eigen-decomposition
More informationGlobal vs. Multiscale Approaches
Harmonic Analysis on Graphs Global vs. Multiscale Approaches Weizmann Institute of Science, Rehovot, Israel July 2011 Joint work with Matan Gavish (WIS/Stanford), Ronald Coifman (Yale), ICML 10' Challenge:
More informationSpectral Clustering. Zitao Liu
Spectral Clustering Zitao Liu Agenda Brief Clustering Review Similarity Graph Graph Laplacian Spectral Clustering Algorithm Graph Cut Point of View Random Walk Point of View Perturbation Theory Point of
More informationData Analysis and Manifold Learning Lecture 3: Graphs, Graph Matrices, and Graph Embeddings
Data Analysis and Manifold Learning Lecture 3: Graphs, Graph Matrices, and Graph Embeddings Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline
More informationStatistical Machine Learning
Statistical Machine Learning Christoph Lampert Spring Semester 2015/2016 // Lecture 12 1 / 36 Unsupervised Learning Dimensionality Reduction 2 / 36 Dimensionality Reduction Given: data X = {x 1,..., x
More informationData Analysis and Manifold Learning Lecture 9: Diffusion on Manifolds and on Graphs
Data Analysis and Manifold Learning Lecture 9: Diffusion on Manifolds and on Graphs Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline of Lecture
More informationInverse problems, Dictionary based Signal Models and Compressed Sensing
Inverse problems, Dictionary based Signal Models and Compressed Sensing Rémi Gribonval METISS project-team (audio signal processing, speech recognition, source separation) INRIA, Rennes, France Ecole d
More informationSpectral Techniques for Clustering
Nicola Rebagliati 1/54 Spectral Techniques for Clustering Nicola Rebagliati 29 April, 2010 Nicola Rebagliati 2/54 Thesis Outline 1 2 Data Representation for Clustering Setting Data Representation and Methods
More informationDATA defined on network-like structures are encountered
Graph Fourier Transform based on Directed Laplacian Rahul Singh, Abhishek Chakraborty, Graduate Student Member, IEEE, and B. S. Manoj, Senior Member, IEEE arxiv:6.v [cs.it] Jan 6 Abstract In this paper,
More informationThe non-backtracking operator
The non-backtracking operator Florent Krzakala LPS, Ecole Normale Supérieure in collaboration with Paris: L. Zdeborova, A. Saade Rome: A. Decelle Würzburg: J. Reichardt Santa Fe: C. Moore, P. Zhang Berkeley:
More informationData-dependent representations: Laplacian Eigenmaps
Data-dependent representations: Laplacian Eigenmaps November 4, 2015 Data Organization and Manifold Learning There are many techniques for Data Organization and Manifold Learning, e.g., Principal Component
More informationINTERPOLATION OF GRAPH SIGNALS USING SHIFT-INVARIANT GRAPH FILTERS
INTERPOLATION OF GRAPH SIGNALS USING SHIFT-INVARIANT GRAPH FILTERS Santiago Segarra, Antonio G. Marques, Geert Leus, Alejandro Ribeiro University of Pennsylvania, Dept. of Electrical and Systems Eng.,
More informationCertifying the Global Optimality of Graph Cuts via Semidefinite Programming: A Theoretic Guarantee for Spectral Clustering
Certifying the Global Optimality of Graph Cuts via Semidefinite Programming: A Theoretic Guarantee for Spectral Clustering Shuyang Ling Courant Institute of Mathematical Sciences, NYU Aug 13, 2018 Joint
More informationGraph Partitioning Using Random Walks
Graph Partitioning Using Random Walks A Convex Optimization Perspective Lorenzo Orecchia Computer Science Why Spectral Algorithms for Graph Problems in practice? Simple to implement Can exploit very efficient
More informationLECTURE NOTE #11 PROF. ALAN YUILLE
LECTURE NOTE #11 PROF. ALAN YUILLE 1. NonLinear Dimension Reduction Spectral Methods. The basic idea is to assume that the data lies on a manifold/surface in D-dimensional space, see figure (1) Perform
More informationSpectral Clustering on Handwritten Digits Database
University of Maryland-College Park Advance Scientific Computing I,II Spectral Clustering on Handwritten Digits Database Author: Danielle Middlebrooks Dmiddle1@math.umd.edu Second year AMSC Student Advisor:
More informationGraphs in Machine Learning
Graphs in Machine Learning Michal Valko INRIA Lille - Nord Europe, France Partially based on material by: Ulrike von Luxburg, Gary Miller, Doyle & Schnell, Daniel Spielman January 27, 2015 MVA 2014/2015
More informationNetwork Topology Inference from Non-stationary Graph Signals
Network Topology Inference from Non-stationary Graph Signals Rasoul Shafipour Dept. of Electrical and Computer Engineering University of Rochester rshafipo@ece.rochester.edu http://www.ece.rochester.edu/~rshafipo/
More informationSpectral Algorithms I. Slides based on Spectral Mesh Processing Siggraph 2010 course
Spectral Algorithms I Slides based on Spectral Mesh Processing Siggraph 2010 course Why Spectral? A different way to look at functions on a domain Why Spectral? Better representations lead to simpler solutions
More informationLaplacian Eigenmaps for Dimensionality Reduction and Data Representation
Laplacian Eigenmaps for Dimensionality Reduction and Data Representation Neural Computation, June 2003; 15 (6):1373-1396 Presentation for CSE291 sp07 M. Belkin 1 P. Niyogi 2 1 University of Chicago, Department
More informationCMPSCI 791BB: Advanced ML: Laplacian Learning
CMPSCI 791BB: Advanced ML: Laplacian Learning Sridhar Mahadevan Outline! Spectral graph operators! Combinatorial graph Laplacian! Normalized graph Laplacian! Random walks! Machine learning on graphs! Clustering!
More informationConvergence Rates of Kernel Quadrature Rules
Convergence Rates of Kernel Quadrature Rules Francis Bach INRIA - Ecole Normale Supérieure, Paris, France ÉCOLE NORMALE SUPÉRIEURE NIPS workshop on probabilistic integration - Dec. 2015 Outline Introduction
More informationMultiscale Wavelets on Trees, Graphs and High Dimensional Data
Multiscale Wavelets on Trees, Graphs and High Dimensional Data ICML 2010, Haifa Matan Gavish (Weizmann/Stanford) Boaz Nadler (Weizmann) Ronald Coifman (Yale) Boaz Nadler Ronald Coifman Motto... the relationships
More informationWeighted Nonlocal Laplacian on Interpolation from Sparse Data
Noname manuscript No. (will be inserted by the editor) Weighted Nonlocal Laplacian on Interpolation from Sparse Data Zuoqiang Shi Stanley Osher Wei Zhu Received: date / Accepted: date Abstract Inspired
More informationGraph Signal Processing
Graph Signal Processing Rahul Singh Data Science Reading Group Iowa State University March, 07 Outline Graph Signal Processing Background Graph Signal Processing Frameworks Laplacian Based Discrete Signal
More informationMachine Learning: Basis and Wavelet 김화평 (CSE ) Medical Image computing lab 서진근교수연구실 Haar DWT in 2 levels
Machine Learning: Basis and Wavelet 32 157 146 204 + + + + + - + - 김화평 (CSE ) Medical Image computing lab 서진근교수연구실 7 22 38 191 17 83 188 211 71 167 194 207 135 46 40-17 18 42 20 44 31 7 13-32 + + - - +
More informationValue Function Approximation with Diffusion Wavelets and Laplacian Eigenfunctions
Value Function Approximation with Diffusion Wavelets and Laplacian Eigenfunctions Sridhar Mahadevan Department of Computer Science University of Massachusetts Amherst, MA 13 mahadeva@cs.umass.edu Mauro
More informationSPECTRAL ANOMALY DETECTION USING GRAPH-BASED FILTERING FOR WIRELESS SENSOR NETWORKS. Hilmi E. Egilmez and Antonio Ortega
SPECTRAL ANOMALY DETECTION USING GRAPH-BASED FILTERING FOR WIRELESS SENSOR NETWORKS Hilmi E. Egilmez and Antonio Ortega Signal and Image Processing Institute, University of Southern California hegilmez@usc.edu,
More informationKernel Learning via Random Fourier Representations
Kernel Learning via Random Fourier Representations L. Law, M. Mider, X. Miscouridou, S. Ip, A. Wang Module 5: Machine Learning L. Law, M. Mider, X. Miscouridou, S. Ip, A. Wang Kernel Learning via Random
More informationLearning gradients: prescriptive models
Department of Statistical Science Institute for Genome Sciences & Policy Department of Computer Science Duke University May 11, 2007 Relevant papers Learning Coordinate Covariances via Gradients. Sayan
More informationCS6220: DATA MINING TECHNIQUES
CS6220: DATA MINING TECHNIQUES Mining Graph/Network Data Instructor: Yizhou Sun yzsun@ccs.neu.edu March 16, 2016 Methods to Learn Classification Clustering Frequent Pattern Mining Matrix Data Decision
More informationTHE HIDDEN CONVEXITY OF SPECTRAL CLUSTERING
THE HIDDEN CONVEXITY OF SPECTRAL CLUSTERING Luis Rademacher, Ohio State University, Computer Science and Engineering. Joint work with Mikhail Belkin and James Voss This talk A new approach to multi-way
More information1 Matrix notation and preliminaries from spectral graph theory
Graph clustering (or community detection or graph partitioning) is one of the most studied problems in network analysis. One reason for this is that there are a variety of ways to define a cluster or community.
More informationLearning from Labeled and Unlabeled Data: Semi-supervised Learning and Ranking p. 1/31
Learning from Labeled and Unlabeled Data: Semi-supervised Learning and Ranking Dengyong Zhou zhou@tuebingen.mpg.de Dept. Schölkopf, Max Planck Institute for Biological Cybernetics, Germany Learning from
More informationTUM 2016 Class 3 Large scale learning by regularization
TUM 2016 Class 3 Large scale learning by regularization Lorenzo Rosasco UNIGE-MIT-IIT July 25, 2016 Learning problem Solve min w E(w), E(w) = dρ(x, y)l(w x, y) given (x 1, y 1 ),..., (x n, y n ) Beyond
More informationProbabilistic & Unsupervised Learning
Probabilistic & Unsupervised Learning Week 2: Latent Variable Models Maneesh Sahani maneesh@gatsby.ucl.ac.uk Gatsby Computational Neuroscience Unit, and MSc ML/CSML, Dept Computer Science University College
More informationProvable Alternating Minimization Methods for Non-convex Optimization
Provable Alternating Minimization Methods for Non-convex Optimization Prateek Jain Microsoft Research, India Joint work with Praneeth Netrapalli, Sujay Sanghavi, Alekh Agarwal, Animashree Anandkumar, Rashish
More informationCS249: ADVANCED DATA MINING
CS249: ADVANCED DATA MINING Graph and Network Instructor: Yizhou Sun yzsun@cs.ucla.edu May 31, 2017 Methods Learnt Classification Clustering Vector Data Text Data Recommender System Decision Tree; Naïve
More informationGraph Functional Methods for Climate Partitioning
Graph Functional Methods for Climate Partitioning Mathilde Mougeot - with D. Picard, V. Lefieux*, M. Marchand* Université Paris Diderot, France *Réseau Transport Electrique (RTE) Buenos Aires, 2015 Mathilde
More informationConjugate gradient acceleration of non-linear smoothing filters
MITSUBISHI ELECTRIC RESEARCH LABORATORIES http://www.merl.com Conjugate gradient acceleration of non-linear smoothing filters Knyazev, A.; Malyshev, A. TR25-43 December 25 Abstract The most efficient edge-preserving
More informationKernels A Machine Learning Overview
Kernels A Machine Learning Overview S.V.N. Vishy Vishwanathan vishy@axiom.anu.edu.au National ICT of Australia and Australian National University Thanks to Alex Smola, Stéphane Canu, Mike Jordan and Peter
More informationHYPERGRAPH BASED SEMI-SUPERVISED LEARNING ALGORITHMS APPLIED TO SPEECH RECOGNITION PROBLEM: A NOVEL APPROACH
HYPERGRAPH BASED SEMI-SUPERVISED LEARNING ALGORITHMS APPLIED TO SPEECH RECOGNITION PROBLEM: A NOVEL APPROACH Hoang Trang 1, Tran Hoang Loc 1 1 Ho Chi Minh City University of Technology-VNU HCM, Ho Chi
More informationSemi-Supervised Learning by Multi-Manifold Separation
Semi-Supervised Learning by Multi-Manifold Separation Xiaojin (Jerry) Zhu Department of Computer Sciences University of Wisconsin Madison Joint work with Andrew Goldberg, Zhiting Xu, Aarti Singh, and Rob
More informationConjugate gradient acceleration of non-linear smoothing filters Iterated edge-preserving smoothing
Cambridge, Massachusetts Conjugate gradient acceleration of non-linear smoothing filters Iterated edge-preserving smoothing Andrew Knyazev (knyazev@merl.com) (speaker) Alexander Malyshev (malyshev@merl.com)
More informationLocally-biased analytics
Locally-biased analytics You have BIG data and want to analyze a small part of it: Solution 1: Cut out small part and use traditional methods Challenge: cutting out may be difficult a priori Solution 2:
More informationReconstruction in the Generalized Stochastic Block Model
Reconstruction in the Generalized Stochastic Block Model Marc Lelarge 1 Laurent Massoulié 2 Jiaming Xu 3 1 INRIA-ENS 2 INRIA-Microsoft Research Joint Centre 3 University of Illinois, Urbana-Champaign GDR
More informationData dependent operators for the spatial-spectral fusion problem
Data dependent operators for the spatial-spectral fusion problem Wien, December 3, 2012 Joint work with: University of Maryland: J. J. Benedetto, J. A. Dobrosotskaya, T. Doster, K. W. Duke, M. Ehler, A.
More informationRecovering any low-rank matrix, provably
Recovering any low-rank matrix, provably Rachel Ward University of Texas at Austin October, 2014 Joint work with Yudong Chen (U.C. Berkeley), Srinadh Bhojanapalli and Sujay Sanghavi (U.T. Austin) Matrix
More informationMATH 829: Introduction to Data Mining and Analysis Clustering II
his lecture is based on U. von Luxburg, A Tutorial on Spectral Clustering, Statistics and Computing, 17 (4), 2007. MATH 829: Introduction to Data Mining and Analysis Clustering II Dominique Guillot Departments
More informationGraphs, Geometry and Semi-supervised Learning
Graphs, Geometry and Semi-supervised Learning Mikhail Belkin The Ohio State University, Dept of Computer Science and Engineering and Dept of Statistics Collaborators: Partha Niyogi, Vikas Sindhwani In
More informationHow to learn from very few examples?
How to learn from very few examples? Dengyong Zhou Department of Empirical Inference Max Planck Institute for Biological Cybernetics Spemannstr. 38, 72076 Tuebingen, Germany Outline Introduction Part A
More informationMachine Learning - MT Clustering
Machine Learning - MT 2016 15. Clustering Varun Kanade University of Oxford November 28, 2016 Announcements No new practical this week All practicals must be signed off in sessions this week Firm Deadline:
More informationCombining geometry and combinatorics
Combining geometry and combinatorics A unified approach to sparse signal recovery Anna C. Gilbert University of Michigan joint work with R. Berinde (MIT), P. Indyk (MIT), H. Karloff (AT&T), M. Strauss
More information1 Matrix notation and preliminaries from spectral graph theory
Graph clustering (or community detection or graph partitioning) is one of the most studied problems in network analysis. One reason for this is that there are a variety of ways to define a cluster or community.
More informationWavelets and Filter Banks on Graphs
Wavelets and Filter Banks on Graphs Pierre Vandergheynst Signal Processing Lab, EPFL Joint work with David Shuman Duke Workshop on Sensing and Analysis of High-Dimensional Data Duke University, July 2011
More informationNonlinear Dimensionality Reduction
Outline Hong Chang Institute of Computing Technology, Chinese Academy of Sciences Machine Learning Methods (Fall 2012) Outline Outline I 1 Kernel PCA 2 Isomap 3 Locally Linear Embedding 4 Laplacian Eigenmap
More informationWavelets on Graphs, an Introduction
Wavelets on Graphs, an Introduction Pierre Vandergheynst and David Shuman Ecole Polytechnique Fédérale de Lausanne (EPFL) Signal Processing Laboratory {pierre.vandergheynst,david.shuman}@epfl.ch Université
More informationLecture 22: More On Compressed Sensing
Lecture 22: More On Compressed Sensing Scribed by Eric Lee, Chengrun Yang, and Sebastian Ament Nov. 2, 207 Recap and Introduction Basis pursuit was the method of recovering the sparsest solution to an
More information8.1 Concentration inequality for Gaussian random matrix (cont d)
MGMT 69: Topics in High-dimensional Data Analysis Falll 26 Lecture 8: Spectral clustering and Laplacian matrices Lecturer: Jiaming Xu Scribe: Hyun-Ju Oh and Taotao He, October 4, 26 Outline Concentration
More informationRelations Between Adjacency And Modularity Graph Partitioning: Principal Component Analysis vs. Modularity Component Analysis
Relations Between Adjacency And Modularity Graph Partitioning: Principal Component Analysis vs. Modularity Component Analysis Hansi Jiang Carl Meyer North Carolina State University October 27, 2015 1 /
More informationDELFT UNIVERSITY OF TECHNOLOGY
DELFT UNIVERSITY OF TECHNOLOGY REPORT -09 Computational and Sensitivity Aspects of Eigenvalue-Based Methods for the Large-Scale Trust-Region Subproblem Marielba Rojas, Bjørn H. Fotland, and Trond Steihaug
More informationAnalysis of Spectral Kernel Design based Semi-supervised Learning
Analysis of Spectral Kernel Design based Semi-supervised Learning Tong Zhang IBM T. J. Watson Research Center Yorktown Heights, NY 10598 Rie Kubota Ando IBM T. J. Watson Research Center Yorktown Heights,
More informationGeometric Modeling Summer Semester 2012 Linear Algebra & Function Spaces
Geometric Modeling Summer Semester 2012 Linear Algebra & Function Spaces (Recap) Announcement Room change: On Thursday, April 26th, room 024 is occupied. The lecture will be moved to room 021, E1 4 (the
More informationLecture: Local Spectral Methods (3 of 4) 20 An optimization perspective on local spectral methods
Stat260/CS294: Spectral Graph Methods Lecture 20-04/07/205 Lecture: Local Spectral Methods (3 of 4) Lecturer: Michael Mahoney Scribe: Michael Mahoney Warning: these notes are still very rough. They provide
More informationDiscrete Signal Processing on Graphs: Sampling Theory
IEEE TRANS. SIGNAL PROCESS. TO APPEAR. 1 Discrete Signal Processing on Graphs: Sampling Theory Siheng Chen, Rohan Varma, Aliaksei Sandryhaila, Jelena Kovačević arxiv:153.543v [cs.it] 8 Aug 15 Abstract
More informationLinear Spectral Hashing
Linear Spectral Hashing Zalán Bodó and Lehel Csató Babeş Bolyai University - Faculty of Mathematics and Computer Science Kogălniceanu 1., 484 Cluj-Napoca - Romania Abstract. assigns binary hash keys to
More informationIntroduction to Machine Learning. PCA and Spectral Clustering. Introduction to Machine Learning, Slides: Eran Halperin
1 Introduction to Machine Learning PCA and Spectral Clustering Introduction to Machine Learning, 2013-14 Slides: Eran Halperin Singular Value Decomposition (SVD) The singular value decomposition (SVD)
More informationMachine Learning for Data Science (CS4786) Lecture 11
Machine Learning for Data Science (CS4786) Lecture 11 Spectral clustering Course Webpage : http://www.cs.cornell.edu/courses/cs4786/2016sp/ ANNOUNCEMENT 1 Assignment P1 the Diagnostic assignment 1 will
More informationCommunities, Spectral Clustering, and Random Walks
Communities, Spectral Clustering, and Random Walks David Bindel Department of Computer Science Cornell University 26 Sep 2011 20 21 19 16 22 28 17 18 29 26 27 30 23 1 25 5 8 24 2 4 14 3 9 13 15 11 10 12
More informationPCA and admixture models
PCA and admixture models CM226: Machine Learning for Bioinformatics. Fall 2016 Sriram Sankararaman Acknowledgments: Fei Sha, Ameet Talwalkar, Alkes Price PCA and admixture models 1 / 57 Announcements HW1
More informationPreprocessing & dimensionality reduction
Introduction to Data Mining Preprocessing & dimensionality reduction CPSC/AMTH 445a/545a Guy Wolf guy.wolf@yale.edu Yale University Fall 2016 CPSC 445 (Guy Wolf) Dimensionality reduction Yale - Fall 2016
More informationBeyond Scalar Affinities for Network Analysis or Vector Diffusion Maps and the Connection Laplacian
Beyond Scalar Affinities for Network Analysis or Vector Diffusion Maps and the Connection Laplacian Amit Singer Princeton University Department of Mathematics and Program in Applied and Computational Mathematics
More informationCOMS 4721: Machine Learning for Data Science Lecture 19, 4/6/2017
COMS 4721: Machine Learning for Data Science Lecture 19, 4/6/2017 Prof. John Paisley Department of Electrical Engineering & Data Science Institute Columbia University PRINCIPAL COMPONENT ANALYSIS DIMENSIONALITY
More informationDesign of graph filters and filterbanks
Design of graph filters and filterbanks Nicolas Tremblay, Paulo Gonçalves, Pierre Borgnat To cite this version: Nicolas Tremblay, Paulo Gonçalves, Pierre Borgnat. Design of graph filters and filterbanks.
More informationA Modified Method Using the Bethe Hessian Matrix to Estimate the Number of Communities
Journal of Advanced Statistics, Vol. 3, No. 2, June 2018 https://dx.doi.org/10.22606/jas.2018.32001 15 A Modified Method Using the Bethe Hessian Matrix to Estimate the Number of Communities Laala Zeyneb
More informationSketched Ridge Regression:
Sketched Ridge Regression: Optimization and Statistical Perspectives Shusen Wang UC Berkeley Alex Gittens RPI Michael Mahoney UC Berkeley Overview Ridge Regression min w f w = 1 n Xw y + γ w Over-determined:
More informationLearning Eigenfunctions: Links with Spectral Clustering and Kernel PCA
Learning Eigenfunctions: Links with Spectral Clustering and Kernel PCA Yoshua Bengio Pascal Vincent Jean-François Paiement University of Montreal April 2, Snowbird Learning 2003 Learning Modal Structures
More informationSpectral Feature Selection for Supervised and Unsupervised Learning
Spectral Feature Selection for Supervised and Unsupervised Learning Zheng Zhao Huan Liu Department of Computer Science and Engineering, Arizona State University zhaozheng@asu.edu huan.liu@asu.edu Abstract
More informationPredicting Graph Labels using Perceptron. Shuang Song
Predicting Graph Labels using Perceptron Shuang Song shs037@eng.ucsd.edu Online learning over graphs M. Herbster, M. Pontil, and L. Wainer, Proc. 22nd Int. Conf. Machine Learning (ICML'05), 2005 Prediction
More informationA Local Non-Negative Pursuit Method for Intrinsic Manifold Structure Preservation
Proceedings of the Twenty-Eighth AAAI Conference on Artificial Intelligence A Local Non-Negative Pursuit Method for Intrinsic Manifold Structure Preservation Dongdong Chen and Jian Cheng Lv and Zhang Yi
More informationCheng Soon Ong & Christian Walder. Canberra February June 2018
Cheng Soon Ong & Christian Walder Research Group and College of Engineering and Computer Science Canberra February June 2018 Outlines Overview Introduction Linear Algebra Probability Linear Regression
More informationData Mining Techniques
Data Mining Techniques CS 622 - Section 2 - Spring 27 Pre-final Review Jan-Willem van de Meent Feedback Feedback https://goo.gl/er7eo8 (also posted on Piazza) Also, please fill out your TRACE evaluations!
More informationLaplacian Eigenmaps for Dimensionality Reduction and Data Representation
Introduction and Data Representation Mikhail Belkin & Partha Niyogi Department of Electrical Engieering University of Minnesota Mar 21, 2017 1/22 Outline Introduction 1 Introduction 2 3 4 Connections to
More information