nips nips2009 nips2009-138 knowledge-graph by maker-knowledge-mining

138 nips-2009-Learning with Compressible Priors

Source: pdf

Author: Volkan Cevher

Abstract: We describe a set of probability distributions, dubbed compressible priors, whose independent and identically distributed (iid) realizations result in p-compressible signals. A signal x ∈ RN is called p-compressible with magnitude R if its sorted coefﬁcients exhibit a power-law decay as |x|(i) R · i−d , where the decay rate d is equal to 1/p. p-compressible signals live close to K-sparse signals (K N) in the r -norm (r > p) since their best K-sparse approximation error decreases with O R · K 1/r−1/p . We show that the membership of generalized Pareto, Student’s t, log-normal, Fr´ chet, and log-logistic distributions to the set of compresse ible priors depends only on the distribution parameters and is independent of N . In contrast, we demonstrate that the membership of the generalized Gaussian distribution (GGD) depends both on the signal dimension and the GGD parameters: the expected decay rate of N -sample iid realizations from the GGD with the shape parameter q is given by 1/ [q log (N/q)]. As stylized examples, we show via experiments that the wavelet coefﬁcients of natural images are 1.67-compressible whereas their pixel gradients are 0.95 log (N/0.95)-compressible, on the average. We also leverage the connections between compressible priors and sparse signals to develop new iterative re-weighted sparse signal recovery algorithms that outperform the standard 1 -norm minimization. Finally, we describe how to learn the hyperparameters of compressible priors in underdetermined regression problems by exploiting the geometry of their order statistics during signal recovery. 1

Reference: text

Summary: the most important sentenses genereted by tfidf model

sentIndex sentText sentNum sentScore

1 edu Abstract We describe a set of probability distributions, dubbed compressible priors, whose independent and identically distributed (iid) realizations result in p-compressible signals. [sent-2, score-0.68]

2 A signal x ∈ RN is called p-compressible with magnitude R if its sorted coefﬁcients exhibit a power-law decay as |x|(i) R · i−d , where the decay rate d is equal to 1/p. [sent-3, score-0.396]

3 p-compressible signals live close to K-sparse signals (K N) in the r -norm (r > p) since their best K-sparse approximation error decreases with O R · K 1/r−1/p . [sent-4, score-0.322]

4 We show that the membership of generalized Pareto, Student’s t, log-normal, Fr´ chet, and log-logistic distributions to the set of compresse ible priors depends only on the distribution parameters and is independent of N . [sent-5, score-0.196]

5 In contrast, we demonstrate that the membership of the generalized Gaussian distribution (GGD) depends both on the signal dimension and the GGD parameters: the expected decay rate of N -sample iid realizations from the GGD with the shape parameter q is given by 1/ [q log (N/q)]. [sent-6, score-0.627]

6 As stylized examples, we show via experiments that the wavelet coefﬁcients of natural images are 1. [sent-7, score-0.32]

7 We also leverage the connections between compressible priors and sparse signals to develop new iterative re-weighted sparse signal recovery algorithms that outperform the standard 1 -norm minimization. [sent-11, score-1.107]

8 Finally, we describe how to learn the hyperparameters of compressible priors in underdetermined regression problems by exploiting the geometry of their order statistics during signal recovery. [sent-12, score-0.818]

9 1 Introduction Many problems in signal processing, machine learning, and communications can be cast as a linear regression problem where an unknown signal x ∈ RN is related to its observations y ∈ RM via y = Φx + n. [sent-13, score-0.294]

10 For instance, CS and SBL frameworks assume that the signal x belongs to the set of sparse signals. [sent-21, score-0.198]

11 CS and SLB algorithms then regularize the solution space by signal priors that promote sparseness and they have been extremely successful in practice in a number of applications even if M N [1–3]. [sent-23, score-0.243]

12 Two more key ingredients are required: (i) the observation matrix Φ must stably embed (or encode) the set of signals x into the space of y, and (ii) a tractable decoding algorithm must exist to map y back to x. [sent-25, score-0.284]

13 Tractable decoding is important for practical reasons as we have limited time and resources, and it can clearly restrict the class of usable signal priors. [sent-28, score-0.239]

14 In this paper, we describe compressible prior distributions whose independent and identically distributed (iid) realizations result in compressible signals. [sent-29, score-1.286]

15 A signal is compressible when sorted magnitudes of its coefﬁcients exhibit a power-law decay. [sent-30, score-0.751]

16 For certain decay rates, compressible signals live close to the sparse signals, i. [sent-31, score-0.842]

17 It is wellknown that the set of K-sparse signals has stable and tractable encoder-decoder pairs (Φ, ∆) for M as small as O(K log (N/K)) [1, 5]. [sent-34, score-0.193]

18 Hence, an N -dimensional compressible signal with the proper decay rate inherits the encoder-decoder pairs of its K-sparse approximation for a given approximation error, and can be stably embedded into dimensions logarithmic in N . [sent-35, score-0.885]

19 Compressible priors analytically summarize the set of compressible signals and shed new light on underdetermined linear regression problems by building upon the literature on sparse signal recovery. [sent-36, score-1.004]

20 Our main results are summarized as follows: 1) By using order statistics, we show that the compressibility of the iid realizations of generalized Pareto, Student’s t, Fr´ chet, and log-logistics distributions is independent of the signals’ dimension. [sent-37, score-0.476]

21 e These distributions are natural members of compressible priors: they truly support logarithmic dimensionality reduction and have important parameter learning guarantees from ﬁnite sample sizes. [sent-38, score-0.68]

22 We demonstrate that probabilistic models for the wavelet coefﬁcients of natural images must also be a natural member of compressible priors. [sent-39, score-0.898]

23 2) We point out a common misconception about the generalized Gaussian distribution (GGD): GGD generates signals that lose their compressibility as N grows. [sent-40, score-0.301]

24 We show that signals generated from Laplacian distribution can only be stably embedded into lower dimensions that grow proportional to N . [sent-44, score-0.199]

25 3) We use compressible priors as a scaffold to build new decoding algorithms based on Bayesian inference arguments. [sent-46, score-0.735]

26 The objective of these algorithms is to approximate the signal realization from a compressible prior as opposed to pragmatically producing sparse solutions. [sent-47, score-0.798]

27 We show how the tuning of these algorithms explicitly depends on the compressible prior parameters, and how to learn the parameters of the signal’s compressible prior on the ﬂy while recovering the signal. [sent-49, score-1.16]

28 Section 2 provides the necessary background on sparse signal recovery. [sent-51, score-0.198]

29 Section 3 mathematically describes the compressible signals and ties them with the order statistics of distributions to introduce compressible priors. [sent-52, score-1.255]

30 Section 4 deﬁnes compressible priors, identiﬁes common misconceptions about the GGD distribution, and examines natural images as instances of compressible priors. [sent-53, score-1.191]

31 Section 6 describes an algorithm for learning the parameters of compressible priors. [sent-55, score-0.547]

32 To account for sparse signals in an appropriate basis, (1) should be modiﬁed as y = Φx + n = ΦΨα + n. [sent-59, score-0.186]

33 Moreover, RIP implies the recovery of K-sparse signals to within a given error bound, and the best attainable lower bounds for M are related to the Gelfand width of ΣK , which is logarithmic in the signal dimension, i. [sent-62, score-0.356]

34 Without loss of generality, we restrict our attention in the sequel to canonically sparse signals and assume that Ψ = I (the N × N identity matrix) so that x = α. [sent-65, score-0.186]

35 Deﬁning xi = |xi |, we arrange the signal coefﬁcients xi in ¯ decreasing order of magnitude as x(1) ≥ x(2) ≥ . [sent-78, score-0.237]

36 ¯ Signals in sw p (R) can be well-approximated by sparse signals as the best K-term approximation error decays rapidly to zero as x − xK r (r/p − 1)−1/r RK 1/r−1/p , when p < r. [sent-85, score-0.326]

37 Suppose now the individual entries xi of the signal x are random variables (RV) drawn iid with respect to a probability density function (pdf) f (x), i. [sent-92, score-0.362]

38 ¯ ¯ The concept of OS enables us to create a link between signals summarized by pdf’s and their compressibility, which is a deterministic property after the signals are realized. [sent-100, score-0.296]

39 F A well-known quantile approximation to the expected OS of a pdf is given by [10]: ¯ E[¯(i) ] = F x 1− i N +1 , (7) where E[·] is the expected value. [sent-105, score-0.191]

40 In practice, the deviations are relatively small for compressible priors. [sent-108, score-0.547]

41 In Sections 4–6, we will use the quantile approximation in (7) as our basis to motivate the set of compressible priors, derive recovery algorithms for x, and learn the parameters of compressible priors during recovery. [sent-109, score-1.335]

42 (9) Table 4 lists example pdf’s, parameterized by θ = (q, λ) 0, and the sw p (R) parameters of their N -sample iid realizations. [sent-111, score-0.299]

43 Section 3); hence, the example pdf’s are compressible priors whenever p < 1. [sent-113, score-0.643]

44 In (9), we make it explicit that the sw p (R) parameters can depend on the parameters θ of the speciﬁc compressible prior as well as the signal dimension N . [sent-114, score-0.842]

45 The dependence of the parameter p on N is of particular interest since it has important implications in signal recovery as well as parameter learning from ﬁnite sample sizes, as discussed below. [sent-115, score-0.195]

46 We deﬁne natural p-compressible priors as the set Np of compressible priors such that p = p(θ) < 1 is independent of N , ∀f (x; θ) ∈ Np . [sent-116, score-0.774]

47 It is possible to prove that we can capture most of the 1 energy in an N -sample iid realization from a natural p-compressible prior by using a constant K, p i. [sent-117, score-0.272]

48 Hence, N -sample iid signal realizations from the compressible priors in Np can be truly embedded into dimensions M that grow logarithmically with N with tractable decoding guarantees due to (3). [sent-120, score-1.346]

49 e It then only comes as a surprise that generalized Gaussian distribution (GGD) is not a natural pcompressible prior since its iid realizations lose their compressibility as N grows (cf. [sent-122, score-0.578]

50 While it is common practice to use a GGD prior with q ≤ 1 for sparse signal recovery, we have no recovery guarantees for signals generated from GGD when M grows logarithmically with N in (1). [sent-124, score-0.484]

51 Along with GGD, Table 4 shows how Weibull, gamma, and lognormal distributions are dimension-restricted in their membership to the set of compressible priors. [sent-127, score-0.596]

52 Unsurprisingly, wavelet thresholding is proven nearoptimal for representing and denoising Besov space images. [sent-131, score-0.196]

53 As the simplest example, the magnitude sorted discrete wavelet coefﬁcients w(i) of a Besov q-image should satisfy w(i) = R · i−1/q . [sent-132, score-0.257]

54 2 Here, we assume that the reader is familiar with the discrete wavelet transform and its properties [11]. [sent-138, score-0.196]

55 4 coefﬁcients while trying to simultaneously capture the dependencies observed in the marginal and joint distributions of natural image wavelet coefﬁcients. [sent-139, score-0.283]

56 Probabilistic approaches are quite important in image compression because optimal compressors quantize the wavelet coefﬁcients according to the estimated distributions, dictating the image compression limits via Shannon’s coding theorem. [sent-140, score-0.248]

57 We conjecture that probabilistic models that summarize the wavelet coefﬁcients of natural images belong to the set of natural (non-iid) p-compressible priors. [sent-141, score-0.351]

58 We base our claim on two observations: 1) Due to the multiscale nature of the wavelet transform, the decay proﬁle of the magnitude sorted wavelet coefﬁcients are scale-invariant, i. [sent-142, score-0.535]

59 Hence, probabilistic models that explain the wavelet transform of any signals should exhibit this decay proﬁle inheritance property. [sent-145, score-0.482]

60 2) The magnitude sorted wavelet coefﬁcients of natural images exhibit a constant decay rate, as expected of Besov space images. [sent-146, score-0.46]

61 2 demonstrates the ideas using natural images from the Berkeley natural images database. [sent-148, score-0.194]

62 5 Signal Decoding Algorithms Convex problems to recover sparse or compressible signals in (2) are usually motivated by Bayesian inference. [sent-149, score-0.733]

63 In a similar fashion, we formalize two new decoding algorithms below by assuming prior distributions on the signal x and the noise n, and then asking inference questions given y in (1). [sent-150, score-0.298]

64 1 Fixed point continuation for a non-iid compressible prior The multivariate Lomax distribution (MLD) provides an elementary example of a non-iid compress−q−N ible prior. [sent-152, score-0.64]

65 In the sequel, we assume λi = λ ∀i, for which i=1 i it can be proved that MLD is compressible with p = 1 [14]. [sent-158, score-0.547]

66 2) MAP Decoder: Suppose that the noise coefﬁcients (ni ’s in (1)) are iid Gaussian with zero mean and variance σ 2 , ni ∼ N (n; 0, σ 2 ). [sent-168, score-0.184]

67 2 Iterative s -decoding for iid scale mixtures of GGD We consider a generalization of GPD and the Student’s t distribution, which we will denote as the generalized Gaussian gamma scale mixture distribution (SMD, in short), whose pdf is given −(q+1)/s s by SMD(x; q, λ, s) ∝ (1 + |x| /λs ) . [sent-177, score-0.357]

68 We assume that x is an N sample iid realization from SMD(x; q, λ, s) with known parameters (q, λ, s) 0 and choose a solution x that maximizes the SMD likehood to ﬁnd the true vector x among the kernel of Φ: x = max SMD(x; q, λ, s) = min x x log 1 + λ−s |xi |s , s. [sent-189, score-0.238]

69 Parameter Learning for Compressible Distributions While deriving decoding algorithms in Section 5, we assumed that the signal coefﬁcients xi are generated from a compressible prior f (x; θ) and that θ is known. [sent-197, score-0.85]

70 To see this, we ﬁrst note that the recovered signals x in general consist of M N non-zero coefﬁcients that mimic the best K-term approximation of the signal xK and some other coefﬁcients that explain the small tail energy. [sent-206, score-0.307]

71 One way to overcome this dependency issue is to treat the recovered signals as if they are drawn iid from a censored GPD. [sent-209, score-0.319]

72 As an alternative, we propose to exploit geometry and use the consensus among the coefﬁcients in ﬁtting the sw p (R) parameters via the auxiliary signal estimates x{k} during iterative recovery. [sent-211, score-0.318]

73 To do this, we employ Fischler and Bolles’ probabilistic random sampling consensus (RANSAC) algorithm [22] to ﬁt a line, whose y-intercept is log R(N, θ) and whose slope is 1/p(N, θ): log xi,{k} = log R(N, θ) − 1 log i, for i = 1, . [sent-212, score-0.251]

74 We determine this threshold by bounding the tail probability that the OS of a compressible prior will be out of bounds. [sent-220, score-0.58]

75 1 Order Statistics To demonstrate the sw p (R) decay proﬁle of p-compressible priors, we generated iid realizations of GGD with q = 1 (LD) and GPD with q = 1, and (non-iid) realizations of MLD with q = 1 of varying signal dimensions N = 10j , where j = 2, 3, 4, 5. [sent-224, score-0.794]

76 To live in sw p (1) with 0 < p ≤ 1, the slope of the resulting curve must be less than or equal to −1. [sent-233, score-0.21]

77 Figure 1(a) illustrates that the iid LD slope is much greater than −1 and moreover logarithmically grows with N . [sent-234, score-0.298]

78 1(b) shows that iid GPD with q = 1 exhibits the constant slope of −1 that is independent of N . [sent-236, score-0.252]

79 The latter two distributions thus produce compressible signal realizations, while the Laplacian does not. [sent-239, score-0.72]

80 2 Natural Images We investigate the images from the Berkeley natural images database in the context of pcompressible priors. [sent-241, score-0.186]

81 , 8), take their wavelet transforms (scaling ﬁlter: daub2), and plot the average of their magnitude ordered wavelet coefﬁcients in Figs. [sent-245, score-0.462]

82 6;λ=10) 1 2 3 4 5 ordered index [power of 10] (a) wavelet coefﬁcients 5 4 3 2 1 0 −5 0 average of 100 images histogram fit − GGD 1 2 3 4 5 ordered index [power of 10] (b) wavelet coefﬁcients coefficient amplitudes 5 4 3 2 1 0 coefficient amplitudes Along with the wavelet coefﬁcients, Fig. [sent-249, score-0.892]

83 ¯ Although wavelet coefﬁcients of natural images do not follow an iid distribution, they exhibit a constant decay rate, which can be well-approximated by an iid GPD distribution. [sent-255, score-0.767]

84 This apparent constant decay rate is well-explained by the decay proﬁle inheritance of the wavelet transform across different resolutions and supports the Besov space assumption used in the deterministic approaches. [sent-256, score-0.439]

85 95;λ=25) 1 2 3 4 5 ordered index [power of 10] (c) pixel gradients Figure 2: Approximation of the order statistics and histograms of natural images with GPD and GGD. [sent-264, score-0.19]

86 2(b) demonstrates the GGD histogram ﬁts to the wavelet coefﬁcients, where the GGD exponent q ∈ [0. [sent-266, score-0.246]

87 Unfortunately, these approaches approximate the wavelet coefﬁ7 0. [sent-274, score-0.196]

88 cients of natural images that have almost no approximation power of the overall image. [sent-291, score-0.294]

89 Figure 2(c) compares the magnitude ordered pixel gradients of the images (solid lines) with the expected OS of GGD (dashed line). [sent-295, score-0.183]

90 From the ﬁgure, it appears that the natural image pixel gradients lose their compressibility as the image dimensions grow, similar to the GGD, Weibull, gamma, and log-normal distributions. [sent-296, score-0.28]

91 3 Iterative 1 Decoding We repeat the compressible signal recovery experiment in Section 3. [sent-300, score-0.742]

92 We ﬁrst randomly sample a signal x ∈ RN (N = 256) where the signal coefﬁcients are iid from the GPD distribution with q = 0. [sent-302, score-0.478]

93 We set M = 128 and draw a random M × N matrix x with iid Gaussian entries to obtain y = Φx. [sent-304, score-0.184]

94 We then decode signals via (13) where maximum iterations is set to 5, with the knowledge of the signal parameters and with learning. [sent-305, score-0.282]

95 8 Conclusions3 Compressible priors create a connection between probabilistic and deterministic models for signal compressibility. [sent-322, score-0.292]

96 We demonstrated that when the p-parameter of a compressible prior is independent of the ambient dimension N , it is possible to have truly logarithmic embedding of its iid signal realizations. [sent-324, score-0.984]

97 Moreover, the learned parameters of such compressible priors are dimension agnostic. [sent-325, score-0.643]

98 In contrast, we showed that when the p-parameter depends on N , we have many restrictions in signal embedding and recovery as well as in parameter learning. [sent-326, score-0.22]

99 We motivated many of the existing sparse signal recovery algorithm as instances of a corresponding compressible prior and discussed parameter learning for these priors from dimensionality reduced data. [sent-328, score-0.922]

100 We hope that the iid compressibility view taken in this paper will pave the way for a better understanding of probabilistic non-iid and structured compressibility models. [sent-329, score-0.425]

similar papers computed by tfidf model

tfidf for this paper:

wordName wordTfidf (topN-words)

[('compressible', 0.547), ('ggd', 0.462), ('gpd', 0.231), ('wavelet', 0.196), ('iid', 0.184), ('mld', 0.177), ('signal', 0.147), ('cients', 0.145), ('signals', 0.135), ('realizations', 0.133), ('coef', 0.126), ('sw', 0.115), ('pdf', 0.115), ('smd', 0.109), ('compressibility', 0.109), ('ransac', 0.098), ('os', 0.098), ('priors', 0.096), ('decoding', 0.092), ('decay', 0.082), ('besov', 0.082), ('slope', 0.068), ('pareto', 0.065), ('decoder', 0.065), ('images', 0.062), ('chet', 0.054), ('sparse', 0.051), ('quantile', 0.051), ('xk', 0.051), ('recovery', 0.048), ('rip', 0.048), ('student', 0.046), ('logarithmically', 0.046), ('cs', 0.043), ('ordered', 0.042), ('bpdn', 0.041), ('ld', 0.039), ('rv', 0.037), ('amplitudes', 0.036), ('sbl', 0.036), ('natural', 0.035), ('log', 0.034), ('gamma', 0.034), ('prior', 0.033), ('lose', 0.033), ('fr', 0.033), ('weibull', 0.033), ('continuation', 0.033), ('stably', 0.033), ('sorted', 0.033), ('iterative', 0.032), ('xi', 0.031), ('grow', 0.031), ('wipf', 0.031), ('resolutions', 0.031), ('pro', 0.03), ('bp', 0.03), ('np', 0.029), ('coefficient', 0.029), ('histogram', 0.028), ('magnitude', 0.028), ('underdetermined', 0.028), ('decoders', 0.027), ('ible', 0.027), ('ive', 0.027), ('lomax', 0.027), ('mqf', 0.027), ('pcompressible', 0.027), ('slb', 0.027), ('stylized', 0.027), ('volkan', 0.027), ('gradients', 0.027), ('power', 0.027), ('live', 0.027), ('laplacian', 0.026), ('distributions', 0.026), ('image', 0.026), ('logarithmic', 0.026), ('deterministic', 0.026), ('embedding', 0.025), ('approximation', 0.025), ('compressive', 0.025), ('reweighted', 0.025), ('guarantees', 0.024), ('generalized', 0.024), ('consensus', 0.024), ('exhibit', 0.024), ('xmap', 0.024), ('disappointing', 0.024), ('pixel', 0.024), ('tractable', 0.024), ('le', 0.024), ('probabilistic', 0.023), ('membership', 0.023), ('exponent', 0.022), ('truly', 0.022), ('inheritance', 0.022), ('pursuit', 0.021), ('basis', 0.021), ('realization', 0.02)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.99999976 138 nips-2009-Learning with Compressible Priors

Author: Volkan Cevher

2 0.11681068 222 nips-2009-Sparse Estimation Using General Likelihoods and Non-Factorial Priors

Author: David P. Wipf, Srikantan S. Nagarajan

Abstract: Finding maximally sparse representations from overcomplete feature dictionaries frequently involves minimizing a cost function composed of a likelihood (or data ﬁt) term and a prior (or penalty function) that favors sparsity. While typically the prior is factorial, here we examine non-factorial alternatives that have a number of desirable properties relevant to sparse estimation and are easily implemented using an efﬁcient and globally-convergent, reweighted 1 -norm minimization procedure. The ﬁrst method under consideration arises from the sparse Bayesian learning (SBL) framework. Although based on a highly non-convex underlying cost function, in the context of canonical sparse estimation problems, we prove uniform superiority of this method over the Lasso in that, (i) it can never do worse, and (ii) for any dictionary and sparsity proﬁle, there will always exist cases where it does better. These results challenge the prevailing reliance on strictly convex penalty functions for ﬁnding sparse solutions. We then derive a new non-factorial variant with similar properties that exhibits further performance improvements in some empirical tests. For both of these methods, as well as traditional factorial analogs, we demonstrate the effectiveness of reweighted 1 -norm algorithms in handling more general sparse estimation problems involving classiﬁcation, group feature selection, and non-negativity constraints. As a byproduct of this development, a rigorous reformulation of sparse Bayesian classiﬁcation (e.g., the relevance vector machine) is derived that, unlike the original, involves no approximation steps and descends a well-deﬁned objective function. 1

3 0.083511032 167 nips-2009-Non-Parametric Bayesian Dictionary Learning for Sparse Image Representations

Author: Mingyuan Zhou, Haojun Chen, Lu Ren, Guillermo Sapiro, Lawrence Carin, John W. Paisley

Abstract: Non-parametric Bayesian techniques are considered for learning dictionaries for sparse image representations, with applications in denoising, inpainting and compressive sensing (CS). The beta process is employed as a prior for learning the dictionary, and this non-parametric method naturally infers an appropriate dictionary size. The Dirichlet process and a probit stick-breaking process are also considered to exploit structure within an image. The proposed method can learn a sparse dictionary in situ; training images may be exploited if available, but they are not required. Further, the noise variance need not be known, and can be nonstationary. Another virtue of the proposed method is that sequential inference can be readily employed, thereby allowing scaling to large images. Several example results are presented, using both Gibbs and variational Bayesian inference, with comparisons to other state-of-the-art approaches. 1

4 0.066283204 67 nips-2009-Directed Regression

Author: Yi-hao Kao, Benjamin V. Roy, Xiang Yan

Abstract: When used to guide decisions, linear regression analysis typically involves estimation of regression coefﬁcients via ordinary least squares and their subsequent use to make decisions. When there are multiple response variables and features do not perfectly capture their relationships, it is beneﬁcial to account for the decision objective when computing regression coefﬁcients. Empirical optimization does so but sacriﬁces performance when features are well-chosen or training data are insufﬁcient. We propose directed regression, an efﬁcient algorithm that combines merits of ordinary least squares and empirical optimization. We demonstrate through a computational study that directed regression can generate signiﬁcant performance gains over either alternative. We also develop a theory that motivates the algorithm. 1

5 0.060762864 157 nips-2009-Multi-Label Prediction via Compressed Sensing

Author: John Langford, Tong Zhang, Daniel J. Hsu, Sham M. Kakade

Abstract: We consider multi-label prediction problems with large output spaces under the assumption of output sparsity – that the target (label) vectors have small support. We develop a general theory for a variant of the popular error correcting output code scheme, using ideas from compressed sensing for exploiting this sparsity. The method can be regarded as a simple reduction from multi-label regression problems to binary regression problems. We show that the number of subproblems need only be logarithmic in the total number of possible labels, making this approach radically more efﬁcient than others. We also state and prove robustness guarantees for this method in the form of regret transform bounds (in general), and also provide a more detailed analysis for the linear prediction setting. 1

6 0.058217436 245 nips-2009-Thresholding Procedures for High Dimensional Variable Selection and Statistical Estimation

7 0.054498281 6 nips-2009-A Biologically Plausible Model for Rapid Natural Scene Identification

8 0.051464807 208 nips-2009-Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Matrices via Convex Optimization

9 0.049184583 62 nips-2009-Correlation Coefficients are Insufficient for Analyzing Spike Count Dependencies

10 0.048649766 225 nips-2009-Sparsistent Learning of Varying-coefficient Models with Structural Changes

11 0.046384342 10 nips-2009-A Gaussian Tree Approximation for Integer Least-Squares

12 0.046221055 228 nips-2009-Speeding up Magnetic Resonance Image Acquisition by Bayesian Multi-Slice Adaptive Compressed Sensing

13 0.045260582 79 nips-2009-Efficient Recovery of Jointly Sparse Vectors

14 0.044047911 137 nips-2009-Learning transport operators for image manifolds

15 0.041852988 93 nips-2009-Fast Image Deconvolution using Hyper-Laplacian Priors

16 0.04144213 55 nips-2009-Compressed Least-Squares Regression

17 0.041221537 20 nips-2009-A unified framework for high-dimensional analysis of $M$-estimators with decomposable regularizers

18 0.041012395 147 nips-2009-Matrix Completion from Noisy Entries

19 0.040894087 169 nips-2009-Nonlinear Learning using Local Coordinate Coding

20 0.040708717 140 nips-2009-Linearly constrained Bayesian matrix factorization for blind source separation

similar papers computed by lsi model

lsi for this paper:

topicId topicWeight

[(0, -0.136), (1, -0.005), (2, 0.013), (3, 0.047), (4, -0.003), (5, -0.012), (6, 0.071), (7, -0.05), (8, 0.12), (9, 0.032), (10, -0.02), (11, 0.002), (12, -0.013), (13, 0.05), (14, -0.003), (15, 0.034), (16, -0.007), (17, 0.062), (18, -0.03), (19, 0.012), (20, -0.011), (21, 0.002), (22, -0.004), (23, 0.092), (24, -0.001), (25, 0.071), (26, -0.042), (27, 0.017), (28, -0.046), (29, 0.036), (30, -0.001), (31, -0.048), (32, -0.044), (33, -0.06), (34, 0.004), (35, 0.038), (36, 0.08), (37, -0.045), (38, -0.08), (39, -0.02), (40, 0.009), (41, -0.026), (42, 0.012), (43, -0.027), (44, 0.027), (45, -0.044), (46, -0.039), (47, -0.106), (48, 0.014), (49, 0.023)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.92112535 138 nips-2009-Learning with Compressible Priors

Author: Volkan Cevher

2 0.69140178 222 nips-2009-Sparse Estimation Using General Likelihoods and Non-Factorial Priors

Author: David P. Wipf, Srikantan S. Nagarajan

3 0.63469344 93 nips-2009-Fast Image Deconvolution using Hyper-Laplacian Priors

Author: Dilip Krishnan, Rob Fergus

Abstract: The heavy-tailed distribution of gradients in natural scenes have proven effective priors for a range of problems such as denoising, deblurring and super-resolution. α These distributions are well modeled by a hyper-Laplacian p(x) ∝ e−k|x| , typically with 0.5 ≤ α ≤ 0.8. However, the use of sparse distributions makes the problem non-convex and impractically slow to solve for multi-megapixel images. In this paper we describe a deconvolution approach that is several orders of magnitude faster than existing techniques that use hyper-Laplacian priors. We adopt an alternating minimization scheme where one of the two phases is a non-convex problem that is separable over pixels. This per-pixel sub-problem may be solved with a lookup table (LUT). Alternatively, for two speciﬁc values of α, 1/2 and 2/3 an analytic solution can be found, by ﬁnding the roots of a cubic and quartic polynomial, respectively. Our approach (using either LUTs or analytic formulae) is able to deconvolve a 1 megapixel image in less than ∼3 seconds, achieving comparable quality to existing methods such as iteratively reweighted least squares (IRLS) that take ∼20 minutes. Furthermore, our method is quite general and can easily be extended to related image processing problems, beyond the deconvolution application demonstrated. 1

4 0.59868711 185 nips-2009-Orthogonal Matching Pursuit From Noisy Random Measurements: A New Analysis

Author: Sundeep Rangan, Alyson K. Fletcher

Abstract: A well-known analysis of Tropp and Gilbert shows that orthogonal matching pursuit (OMP) can recover a k-sparse n-dimensional real vector from m = 4k log(n) noise-free linear measurements obtained through a random Gaussian measurement matrix with a probability that approaches one as n → ∞. This work strengthens this result by showing that a lower number of measurements, m = 2k log(n − k), is in fact sufﬁcient for asymptotic recovery. More generally, when the sparsity level satisﬁes kmin ≤ k ≤ kmax but is unknown, m = 2kmax log(n − kmin ) measurements is sufﬁcient. Furthermore, this number of measurements is also sufﬁcient for detection of the sparsity pattern (support) of the vector with measurement errors provided the signal-to-noise ratio (SNR) scales to inﬁnity. The scaling m = 2k log(n − k) exactly matches the number of measurements required by the more complex lasso method for signal recovery in a similar SNR scaling.

5 0.59810764 79 nips-2009-Efficient Recovery of Jointly Sparse Vectors

Author: Liang Sun, Jun Liu, Jianhui Chen, Jieping Ye

Abstract: We consider the reconstruction of sparse signals in the multiple measurement vector (MMV) model, in which the signal, represented as a matrix, consists of a set of jointly sparse vectors. MMV is an extension of the single measurement vector (SMV) model employed in standard compressive sensing (CS). Recent theoretical studies focus on the convex relaxation of the MMV problem based on the (2, 1)-norm minimization, which is an extension of the well-known 1-norm minimization employed in SMV. However, the resulting convex optimization problem in MMV is signiﬁcantly much more difﬁcult to solve than the one in SMV. Existing algorithms reformulate it as a second-order cone programming (SOCP) or semideﬁnite programming (SDP) problem, which is computationally expensive to solve for problems of moderate size. In this paper, we propose a new (dual) reformulation of the convex optimization problem in MMV and develop an efﬁcient algorithm based on the prox-method. Interestingly, our theoretical analysis reveals the close connection between the proposed reformulation and multiple kernel learning. Our simulation studies demonstrate the scalability of the proposed algorithm.

6 0.5962221 167 nips-2009-Non-Parametric Bayesian Dictionary Learning for Sparse Image Representations

7 0.5632782 104 nips-2009-Group Sparse Coding

8 0.51577753 137 nips-2009-Learning transport operators for image manifolds

9 0.51085281 6 nips-2009-A Biologically Plausible Model for Rapid Natural Scene Identification

10 0.50801474 164 nips-2009-No evidence for active sparsification in the visual cortex

11 0.50582552 157 nips-2009-Multi-Label Prediction via Compressed Sensing

12 0.50469661 228 nips-2009-Speeding up Magnetic Resonance Image Acquisition by Bayesian Multi-Slice Adaptive Compressed Sensing

13 0.47951052 7 nips-2009-A Data-Driven Approach to Modeling Choice

14 0.46830299 206 nips-2009-Riffled Independence for Ranked Data

15 0.46725535 245 nips-2009-Thresholding Procedures for High Dimensional Variable Selection and Statistical Estimation

16 0.46686187 208 nips-2009-Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Matrices via Convex Optimization

17 0.46642584 36 nips-2009-Asymptotic Analysis of MAP Estimation via the Replica Method and Compressed Sensing

18 0.4661276 67 nips-2009-Directed Regression

19 0.45756271 224 nips-2009-Sparse and Locally Constant Gaussian Graphical Models

20 0.45476419 180 nips-2009-On the Convergence of the Concave-Convex Procedure

similar papers computed by lda model

lda for this paper:

topicId topicWeight

[(7, 0.017), (24, 0.049), (25, 0.061), (35, 0.066), (36, 0.093), (39, 0.036), (57, 0.315), (58, 0.098), (61, 0.01), (71, 0.036), (81, 0.021), (86, 0.097), (91, 0.011)]

similar papers list:

simIndex simValue paperId paperTitle

1 0.88384181 88 nips-2009-Extending Phase Mechanism to Differential Motion Opponency for Motion Pop-out

Author: Yicong Meng, Bertram E. Shi

Abstract: We extend the concept of phase tuning, a ubiquitous mechanism among sensory neurons including motion and disparity selective neurons, to the motion contrast detection. We demonstrate that the motion contrast can be detected by phase shifts between motion neuronal responses in different spatial regions. By constructing the differential motion opponency in response to motions in two different spatial regions, varying motion contrasts can be detected, where similar motion is detected by zero phase shifts and differences in motion by non-zero phase shifts. The model can exhibit either enhancement or suppression of responses by either different or similar motion in the surrounding. A primary advantage of the model is that the responses are selective to relative motion instead of absolute motion, which could model neurons found in neurophysiological experiments responsible for motion pop-out detection. 1 In trod u ction Motion discontinuity or motion contrast is an important cue for the pop-out of salient moving objects from contextual backgrounds. Although the neural mechanism underlying the motion pop-out detection is still unknown, the center-surround receptive field (RF) organization is considered as a physiological basis responsible for the pop-out detection. The center-surround RF structure is simple and ubiquitous in cortical cells especially in neurons processing motion and color information. Nakayama and Loomis [1] have predicted the existence of motion selective neurons with antagonistic center-surround receptive field organization in 1974. Recent physiological experiments [2][3] show that neurons with center-surround RFs have been found in both middle temporal (MT) and medial superior temporal (MST) areas related to motion processing. This antagonistic mechanism has been suggested to detect motion segmentation [4], figure/ground segregation [5] and the differentiation of object motion from ego-motion [6]. There are many related works [7]-[12] on motion pop-out detection. Some works [7]-[9] are based on spatio-temporal filtering outputs, but motion neurons are not fully interacted by either only inhibiting similar motion [7] or only enhancing opposite motion [8]. Heeger, et al. [7] proposed a center-surround operator to eliminate the response dependence upon rotational motions. But the Heeger's model only shows a complete center-surround interaction for moving directions. With respect to the surrounding speed effects, the neuronal responses are suppressed by the same speed with the center motion but not enhanced by other speeds. Similar problem existed in [8], which only modeled the suppression of neuronal responses in the classical receptive field (CRF) by similar motions in surrounding regions. Physiological experiments [10][11] show that many neurons in visual cortex are sensitive to the motion contrast rather than depend upon the absolute direction and speed of the object motion. Although pooling over motion neurons tuned to different velocities can eliminate the dependence upon absolute velocities, it is computationally inefficient and still can't give full interactions of both suppression and enhancement by similar and opposite surrounding motions. The model proposed by Dellen, et al. [12] computed differential motion responses directly from complex cells in V1 and didn't utilize responses from direction selective neurons. In this paper, we propose an opponency model which directly responds to differential motions by utilizing the phase shift mechanism. Phase tuning is a ubiquitous mechanism in sensory information processing, including motion, disparity and depth detection. Disparity selective neurons in the visual cortex have been found to detect disparities by adjusting the phase shift between the receptive field organizations in the left and right eyes [13][14]. Motion sensitive cells have been modeled in the similar way as the disparity energy neurons and detect image motions by utilizing the phase shift between the real and imaginary parts of temporal complex valued responses, which are comparable to images to the left and right eyes [15]. Therefore, the differential motion can be modeled by exploring the similarity between images from different spatial regions and from different eyes. The remainder of this paper is organized as following. Section 2 illustrates the phase shift motion energy neurons which estimate image velocities by the phase tuning in the imaginary path of the temporal receptive field responses. In section 3, we extend the concept of phase tuning to the construction of differential motion opponency. The phase difference determines the preferred velocity difference between adjacent areas in retinal images. Section 4 investigates properties of motion pop-out detection by the proposed motion opponency model. Finally, in section 5, we relate our proposed model to the neural mechanism of motion integration and motion segmentation in motion related areas and suggest a possible interpretation for adaptive center-surround interactions observed in biological experiments. 2 Phase Shift Motion Energy Neurons Adelson and Bergen [16] proposed the motion energy model for visual motion perception by measuring spatio-temporal orientations of image sequences in space and time. The motion energy model posits that the responses of direction-selective V1 complex cells can be computed by a combination of two linear spatio-temporal filtering stages, followed by squaring and summation. The motion energy model was extended in [15] to be phase tuned by splitting the complex valued temporal responses into real and imaginary paths and adding a phase shift on the imaginary path. Figure 1(a) demonstrates the schematic diagram of the phase shift motion energy model. Here we assume an input image sequence in two-dimensional space (x, y) and time t. The separable spatio-temporal receptive field ensures the cascade implementation of RF with spatial and temporal filters. Due to the requirement of the causal temporal RF, the phase shift motion energy model didn’t adopt the Gabor filter like the spatial RF. The phase shift spatio-temporal RF is modeled with a complex valued function f ( x, y, t ) = g ( x, y ) ⋅ h ( t , Φ ) , where the spatial and temporal RFs are denoted by g ( x, y ) and h ( t , Φ ) respectively, g ( x, y ) = N ( x, y | 0, C ) exp ( jΩ x x + jΩ y y ) h ( t , Φ ) = hreal ( t ) + exp ( jΦ ) himag ( t ) (1) and C is the covariance matrix of the spatial Gaussian envelope and Φ is the phase tuning of the motion energy neuron. The real and imaginary profiles of the temporal receptive field are Gamma modulated sinusoidal functions with quadrature phases, hreal ( t ) = G ( t | α ,τ ) cos ( Ωt t ) (2) himag ( t ) = G ( t | α ,τ ) sin ( Ωt t ) The envelopes for complex exponentials are functions of Gaussian and Gamma distributions, N ( x, y | 0, C ) = ⎛ x2 y2 exp ⎜ − 2 − 2 ⎜ 2σ x 2σ y 2πσ xσ y ⎝ 1 ⎞ ⎟ ⎟ ⎠ (3) hreal (t ) g ( x, y ) himag (t ) g ( x, y ) (·)2 (·)2 M M M (·)2 (·)2 M M M 2 (·) 2 (·) Vreal V (Φ ) e jΦ Vimag (a) Ev ( Φ max ) (·)2 wc ( x, y ) e jΦmin Ev ( (b) M 0 ) w ( x, y ) c M Ev ( Φ min ) M EΔv ( Θ ) ∫∫∫ K x , y ,Φ e j0 e jΦmin ws ( x, y ) Ks c e ∫∫∫ jΘ ws ( x, y ) e j0 x , y ,Φ ws ( x, y ) wc ( x, y ) M e jΦ max e jΦmax (·)2 M (·)2 M M (·)2 M 2 (·) M M (·)2 (c) Figure 1. (a) shows the diagram of the phase shift motion energy model adapted from [15]. (b) draws the spatiotemporal representation of the phase shift motion energy neuron with the real and imaginary receptive field demonstrated by the two left pictures. (c) illustrates the construction of differential motion opponency with a phase difference Θ from two populations of phase shift motion energy neurons in two spatial areas c and s. To avoid clutter, the space location (x, y) is not explicitly shown in phase tuned motion energies. G (t | α ,τ ) = 1 ⎛ t t α −1 exp ⎜ − Γ(α )τ α ⎝ τ ⎞ ⎟ u (t ) ⎠ (4) where Γ (α ) is the gamma function and u ( t ) is the unit step function. The parameters α and τ determine the temporal RF size. As derived in [15], the motion energy at location (x, y) can be computed by E v ( x, y, Φ ) = S + P cos ( Ψ − Φ ) (5) where S = Vreal 2 + Vimag 2 * P = 2 VrealVimag ( * Ψ = arg VrealVimag (6) ) and complex valued responses in real and imaginary paths are obtained as, Vreal ( x, y, t ) = ∫∫∫ g (ξ , ζ ) h (η ) I ( x − ξ , y − ζ , t − η ) dξ dζ dη real ξ ,ζ ,η Vimag ( x, y, t ) = ∫∫∫ g (ξ , ζ ) h (η ) I ( x − ξ , y − ζ , t − η ) dξ dζ dη ξ ζ η (7) imag , , The superscript * represents the complex conjugation and the phase shift parameter Φ controls the spatio-temporal orientation tuning. To avoid clutter, the spatial location variables x and y for S, P, Ψ, Vreal and Vimag are not explicitly shown in Eq. (5) and (6). Figure 1(b) demonstrates the even and odd profiles of the spatio-temporal RF tuned to a particular phase shift. Θ 0 Θ 0 (a) (b) Figure 2. Two types of differential motion opponency constructions of (a) center-surrounding interaction and (b) left-right interaction. Among cells in area MT with surrounding modulations, 25% of cells are with the antagonistic RF structure in the top row and another 50% of cells have the integrative RF structure as shown in the bottom row. 3 Extending Phase Op p on ency Mechanism to D i f f e r e nt i a l Motion Based on the above phase shift motion energy model, the local image velocity at each spatial location can be represented by a phase shift which leads to the peak response across a population of motion energy neurons. Across regions of different motions, there are clear discontinuities on the estimated velocity map. The motion discontinuities can be detected by edge detectors on the velocity map to segment different motions. However, this algorithm for motion discontinuities detection can’t discriminate between the object motion and uniform motions in contextual backgrounds. Here we propose a phase mechanism to detect differential motions inspired by the disparity energy model and adopt the center-surround inhibition mechanism to pop out the object motion from contextual background motions. The motion differences between different spatial locations can be modeled in the similar way as the disparity model. The motion energies from two neighboring locations are considered as the retinal images to the left and right eyes. Thus, we can construct a differential motion opponency by placing two populations of phase shift motion energy neurons at different spatial locations and the energy EΔv ( Θ ) of the opponency is the squared modulus of the averaged phase shift motion energies over space and phase, E Δv ( Θ ) = ∫∫∫ E ( x, y, Φ ) ⋅ w ( x, y, Φ | Θ ) dxdyd Φ v 2 (8) where w ( x, y, Θ ) is the profile for differential motion opponency and Δv is the velocity difference between the two spatial regions defined by the kernel w ( x, y, Θ ) . Since w ( x, y, Θ ) is intended to implement the functional role of spatial interactions, it is desired to be a separable function in space and phase domain and can be modeled by phase tuned summation of two spatial kernels, w ( x, y, Φ | Θ ) = wc ( x, y ) e jΦ + e jΘ+ jΦ ws ( x, y ) (9) where wc ( x, y ) and ws ( x, y ) are Gaussian kernels of different spatial sizes σ c and σ s , and Θ is the phase difference representing velocity difference between two spatial regions c and s. Substituting Eq. (9) into Eq. (8), the differential motion energy can be reformulated as EΔv ( Θ ) = K c + e jΘ K s 2 (10) 3 3 3 2 2 2 1 1 1 0 0 -1 -1 -2 -2 -2 -3 -3 -3 -3 -3 1 Right Velocity Right Velocity 0.98 0.96 0.94 0.92 0 0.9 0.88 -1 0.86 0.84 0.82 -2 -1 0 1 Left Velocity 2 3 -2 -1 0 1 Left Velocity 2 3 0.8 (a) (b) Figure 3. (a) Phase map and (b) peak magnitude map are obtained from stimuli of two patches of random dots moving with different velocities. The two patches of stimuli are statistically independent but share the same spatial properties: dot size of 2 pixels, dot density of 10% and dot coherence level of 100%. The phase tuned population of motion energy neurons are applied to each patch of random dots with RF parameters: Ωt = 2π/8, Ωt = 2π/16, σx = 5 and τ = 5.5. For each combination of velocities from left and right patches, averaged phase shifts over space and time are computed and so do the magnitudes of peak responses. The unit for velocities is pixels per frame. where Kc = ∫∫∫ E ( x, y, Φ ) exp ( jΦ ) w ( x, y ) dxdyd Φ v,c c x , y ,Φ Ks = ∫∫∫ E ( x, y, Φ ) exp ( jΦ ) w ( x, y ) dxdyd Φ v,s (11) s x, y ,Φ Ev ,c ( x, y, Φ ) and Ev , s ( x, y, Φ ) are phase shift motion energies at location (x, y) and with phase shift Φ. Utilizing the results in Eq. (5) and (6), Eq. (10) and (11) generate similar results, E Δv ( Θ ) = Sopp + Popp cos ( Θopp − Θ ) (12) where Sopp = K c 2 + Ks Popp = 2 K c K s* 2 (13) Θopp = arg ( K c K s* ) According to above derivations, by varying the phase shift Θ between –π and π, the relative motion energy of the differential motion opponency can be modeled as population responses across a population of phase tuned motion opponencies. The response is completely specified by three parameters Sopp , Popp and Θopp . The schematic diagram of this opponency is illustrated in Figure 1(c). The differential motion opponency is constituted by three stages. At the first stage, a population of phase shift motion energy neurons is applied to be selective to different velocities. At the second stage, motion energies from the first stage are weighted by kernels tuned to different spatial locations and phase shifts respectively for both spatial regions and two single differential motion signals in region c and region s are achieved by integrating responses from these two regions over space and phase tuning. Finally, the differential motion energy is computed by the squared modulus of the summation of the integrated motion signal in region c and phase shifted motion signal in region s. The subscripts c and s represent two interacted spatial regions which are not limited to the center and surround regions. The opponency could also be constructed by the neighboring left and right Inhibitive interaction, Θ = π/2 Excitatory interaction, Θ =0 Inhibitory 2 1.6 Responses 1.6 Responses Excitatory 2 1.2 0.8 1.2 0.8 0.4 0.4 0 0 pi/2 pi 3pi/2 Surrouding Direction 0 0 2pi (a) Model by Petkov et al. [8] pi/2 pi 3pi/2 Surrouding Direction (b) Model by Heeger et al. [7] Inhibitory 2 2pi Inhibitory 2 1.6 1.6 Responses Responses 1.2 0.8 1.2 0.8 0.4 0.4 0 0 0 pi/2 pi Surrouding Direction 3pi/2 2pi 0 pi/2 pi Surrouding Direction 3pi/2 2pi (c) (d) Figure 4. Demonstrations of center-surround differential motion opponency, where (a) show the excitation of opposite directions outside the CRF and (b) show the inhibition by surrounding motions in same directions. The center-surround inhibition models by Petkov, et al. [8] and Heeger, et al. [7] are shown in (c) and (d). Responses above 1 indicate enhancement and responses below 1 indicate suppressions. spatial regions. Figure 2 shows two types of structures for the differential motion opponency. In [17], the authors demonstrates that among cells in area MT with surrounding modulations, 25% of cells are with the antagonistic RF structure as shown in Figure 2(a) and another 50% of cells have the integrative RF structure as shown in Figure 2(b). The velocity difference tuning of the opponency is determined by the phase shift parameter Θ combined with parameters of spatial and temporal frequencies for motion energy neurons. The larger phase shift magnitude prefers the bigger velocity difference. This phase tuning of velocity difference is consistent with the phase tuning of motion energy neurons. Figure 3 shows the phase map obtained by using random dots stimuli with different velocities on two spatial patches (left and right patches with sizes of 128 pixels 128 pixels). Along the diagonal line, velocities from left and right patches are equal to each other and therefore phase estimates are zeros along this line. Deviated from the diagonal line to upper-left and lower-right, the phase magnitudes increase while positive phases indicate larger left velocities and negative phases indicate larger right velocities. The phase tuning can give a good classification of velocity differences. 4 V a l i d a t i o n o f D i f f e r e n t i a l M o t i o n O pp o n e n c y Out derivation and analysis above show that the phase shift between two neighboring spatial regions is a good indicator for motion difference between these two regions. In this section, we validate the proposed differential motion opponency by two sets of experiments, which show effects of both surrounding directions and speeds on the center motion. Inhibitory 2 1.6 1.2 1.2 Responses 1.6 Responses Inhibitory 2 0.8 0.4 0.4 0 -2 0.8 0 -1.5 -1 -0.5 0 0.5 Center Speed 1 1.5 2 -2 -1.5 -1 -0.5 0 0.5 Center Speed 1 1.5 2 (a) (b) Figure 5. The insensitivity of the proposed opponency model to absolute center and surrounding velocities is demonstrated in (a), where responses are enhanced for all center velocities from -2 to 2 pixels per frame. In (b), the model by Heeger, et al. [7] only shows enhancement when the center speed matches the preferred speed of 1.2 pixel per frame. Similarly, responses above 1 indicate enhancement and below 1 indicate suppressions. In both curves, the velocity differences between center and surrounding regions are maintained as a constant of 3 pixels per frame. Physiological experiments [2][3] have demonstrated that the neuronal activities in the classical receptive field are suppressed by responses outside the CRF to stimuli with similar motions including both directions and speeds on the center and surrounding regions. On the contrary, visual stimuli of opposite directions or quite different speeds outside the CRF enhance the responses in the CRF. In their experiments, they used a set of stimuli of random dots moving at different velocities, where there are small patches of moving random dots on the center. We tested the properties of the proposed opponency model for motion difference measurement by using similar random dots stimuli. The random dots on background move with different speeds and in different direction but have the same statistical parameters: dot size of 2 pixels, dot density of 10% and motion coherence level of 100%. The small random dots patches are placed on the center of background stimuli to stimulate the neurons in the CRF. These small patches share the same statistical parameters with background random dots but move with a constant velocity of 1 pixel per frame. Figure 4 shows results for the enhanced and suppressed responses in the CRF with varying surrounding directions. The phase shift motion energy neurons had the same spatial and temporal frequencies and the same receptive field sizes, and were selective to vertical orientations. The preferred spatial frequency was 2π/16 radian per pixel and the temporal frequency was 2π/16 radian per frame. The sizes of RF in horizontal and vertical directions were respectively 5 pixels and 10 pixels, corresponding to a spatial bandwidth of 1.96 octaves. The time constant τ was 5.5 frames which resulted in a temporal bandwidth of 1.96 octaves. As shown in Figure 4 (a) and (b), the surrounding motion of opposite direction gives the largest response to the motion in the CRF for the inhibitory interaction and the smallest response for the excitatory interaction. Results demonstrated in Figure 4 are consistent with physiological results reported in [3]. In Born’s paper, inhibitory cells show response enhancement and excitatory cells show response suppression when surrounding motions are in opposite directions. The 3-dB bandwidth for the surrounding moving direction is about 135 degrees for the physiological experiments while the bandwidth is about 180 degrees for the simulation results in our proposed model. Models proposed by Petkov, et al. [8] and Heeger, et al. [7] also show clear inhibition between opposite motions. The Petkov’s model achieves the surrounding suppression for each point in ( x, y, t ) space by the subtraction between responses from that point and its surroundings and followed by a half-wave rectification, + % Ev ,θ ( x, y, t ) = Ev ,θ ( x, y, t ) − α ⋅ Sv ,θ ( x, y, t ) (14) where Ev ,θ ( x, y, t ) is the motion energy at location (x,y) and time t for a given preferred speed v and orientation θ, Sv ,θ ( x, y, t ) is the average motion energy in the surrounding of point (x, y, t), % Ev ,θ ( x, y, t ) is the suppressed motion energy and the factor α controls the inhibition strength. The inhibition term is computed by weighted motion energy Sv ,θ ( x, y, t ) = Ev ,θ ( x, y, t ) ∗ wv ,θ ( x, y, t ) (15) where wv ,θ ( x, y, t ) is the surround weighting function. The Heeger’s model constructs the center-surround motion opponent by computing the weighted sum of responses from motion selective cells, Rv ,θ ( t ) = ∑ β ( x, y ) ⎡ Ev ,θ ( x, y, t ) − E− v ,θ ( x, y, t ) ⎤ ⎣ ⎦ (16) x, y where β ( x, y ) is a center-surround weighting function and the motion energy at each point should be normalized across all cells with different tuning properties. As shown in Figure 4 (c) and (d) for results of Petkov’s and Heeger’s models, we replace the conventional frequency tuned motion energy neuron with our proposed phase tuned neuron. The model by Petkov, et al. [8] is generally suppressive and only reproduces less suppression for opposite motions, which is inconsistent with results from [3]. The model by Heeger, et al. [7] has similar properties with our proposed model with respect to both excitatory and inhibitory interactions. To evaluate the sensitivity of the proposed opponency model to velocity differences, we did simulations by using similar stimuli with the above experiment in Figure 4 but maintaining a constant velocity difference of 3 pixels per frame between the center and surrounding random dot patches. As shown in Figure 5, by varying the velocities of random dots on the center region, we found that responses by the proposed model are always enhanced independent upon absolute velocities of center stimuli, but responses by the Heeger’s model achieve the enhancement at a center velocity of 1.2 pixels per frame and maintain suppressed at other speeds. 5 D i s c u s s i on We proposed a new biologically plausible model of the differential motion opponency to model the spatial interaction property of motion energy neurons. The proposed opponency model is motivated by the phase tuning mechanism of disparity energy neurons which infers the disparity information from the phase difference between complex valued responses to left and right retinal images. Hence, the two neighboring spatial areas can be considered as left and right images and the motion difference between these two spatial regions is detected by the phase difference between the complex valued responses at these two regions. Our experimental results demonstrate a consistent conclusion with physiological experiments that motions of opposite directions and different speeds outside the CRF can show both inhibitive and excitatory effects on the CRF responses. The inhibitive interaction helps to segment the moving object from backgrounds when fed back to low-level features such as edges, orientations and color information. Except providing a unifying phase mechanism in understanding neurons with different functional roles at different brain areas, the proposed opponency model could possibly provide a way to understand the motion integration and motion segmentation. Integration and segmentation are two opposite motion perception tasks but co-exist to constitute two fundamental types of motion processing. Segmentation is achieved by discriminating motion signals from different objects, which is thought to be due to the antagonistic interaction between center and surrounding RFs. Integration is obtained by utilizing the enhancing function of surrounding areas to CRF areas. Both types of processing have been found in motion related areas including area MT and MST. Tadin, et al. [18] have found that motion segmentation dominants at high stimulus contrast and gives the way to motion integration at low stimulus contrast. Huang, et al. [19] suggests that the surrounding modulation is adaptive according to the visual stimulus such as contrasts and noise levels. Since our proposed opponency model determines the functional role of neurons by only the phase shift parameter, this makes the proposed model to be an ideal candidate model for the adaptive surrounding modulation with the phase tuning between two spatial regions. References [1]. K. Nakayama and J. M. Loomis, “Optical velocity patterns, velocity-sensitive neurons, and space perception: A hypothesis,” Perception, vol. 3, 63-80, 1974. [2]. K. Tanaka, K. Hikosaka, H. Saito, M. Yukie, Y. Fukada and E. Iwai, “Analysis of local and wide-field movements in the superior temporal visual areas of the macaque monkey,” Journal of Neuroscience, vol. 6, pp. 134-144, 1986. [3]. R. T. Born and R. B. H. Tootell, “Segregation of global and local motion processing in primate middle temporal visual area,” Nature, vol. 357, pp. 497-499, 1992. [4]. J. Allman, F. Miezin and E. McGuinness, “Stimulus specific responses from beyond the classical receptive field: Neurophysiological mechanisms for local-global comparisions in visual neurons,” Annual Review Neuroscience, vol. 8, pp. 407-430, 1985. [5]. V. A. F. Lamme, “The neurophysiology of figure-ground segregation in primary visual cortex,” Journal of Neuroscience, vol. 15, pp. 1605-1615, 1995. [6]. D. C. Bradley and R. A. Andersen, “Center-surround antagonism based on disparity in primate area MT,” Journal of Neuroscience, vol. 18, pp. 7552-65, 1998. [7]. D. J. Heeger, A. D. Jepson and E. P. Simoncelli, “Recovering observer translation with center-surround operators,” Proc IEEE Workshop on Visual Motion, pp. 95-100, Oct 1991. [8]. N. Petkov and E. Subramanian, “Motion detection, noise reduction, texture suppression, and contour enhancement by spatiotemporal Gabor filters with surround inhibition,” Biological Cybernetics, vol. 97, pp. 423-439, 2007. [9]. M. Escobar and P. Kornprobst, “Action recognition with a Bio-inspired feedforward motion processing model: the richness of center-surround interactions,” ECCV '08: Proceedings of the 10th European Conference on Computer Vision, pp. 186-199, Marseille, France, 2008. [10]. B. J. Frost and K. Nakayama, “Single visual neurons code opposing motion independent of direction,” Science, vol. 200, pp. 744-745, 1983. [11]. A. Cao and P. H. Schiller, “Neural responses to relative speed in the primary visual cortex of rhesus monkey,” Visual Neuroscience, vol. 20, pp. 77-84, 2003. [12]. B. K. Dellen, J. W. Clark and R. Wessel, “Computing relative motion with complex cells,” Visual Neuroscience, vol. 22, pp. 225-236, 2005. [13]. I. Ohzawa, G. C. Deangelis and R. D. Freeman, “Encoding of binocular disparity by complex cells in the cat’s visual cortex,” Journal of Neurophysiology, vol. 77, pp. 2879-2909, 1997. [14]. D. J. Fleet, H. Wagner and D. J. Heeger, “Neural Encoding of binocular disparity: energy model, position shifts and phase shifts,” Vision Research, vol. 26, pp. 1839-1857, 1996. [15]. Y. C. Meng and B. E. Shi, “Normalized Phase Shift Motion Energy Neuron Populations for Image Velocity Estimation,” International Joint Conference on Neural Network, Atlanta, GA, June 14-19, 2009. [16]. E. H. Adelson and J. R. Bergen, “Spatiotemporal energy models for the perception of motion,” J. Opt. Soc. Am. A Opt. Image Sci. Vis., vol. 2, pp. 284-299, 1985. [17]. D. K. Xiao, S. Raiguel, V. Marcar, J. Koenderink and G. A. Orban, “The spatial distribution of the antagonistic surround of MT/V5,” Cereb Cortex, vol. 7, pp. 662-677, 1997. [18]. D. Tadin, J. S. Lappin, L. A. Gilroy and R. Blake, “Perceptual consequences of centre-surround antagonism in visual motion processing,” Nature, vol. 424, pp. 312-315, 2003. [19]. X. Huang, T. D. Albright and G. R. Stoner, “Adaptive surround modulation in cortical area MT,” Neuron, vol. 53, pp. 761-770, 2007.

2 0.8327713 232 nips-2009-Strategy Grafting in Extensive Games

Author: Kevin Waugh, Nolan Bard, Michael Bowling

Abstract: Extensive games are often used to model the interactions of multiple agents within an environment. Much recent work has focused on increasing the size of an extensive game that can be feasibly solved. Despite these improvements, many interesting games are still too large for such techniques. A common approach for computing strategies in these large games is to ﬁrst employ an abstraction technique to reduce the original game to an abstract game that is of a manageable size. This abstract game is then solved and the resulting strategy is played in the original game. Most top programs in recent AAAI Computer Poker Competitions use this approach. The trend in this competition has been that strategies found in larger abstract games tend to beat strategies found in smaller abstract games. These larger abstract games have more expressive strategy spaces and therefore contain better strategies. In this paper we present a new method for computing strategies in large games. This method allows us to compute more expressive strategies without increasing the size of abstract games that we are required to solve. We demonstrate the power of the approach experimentally in both small and large games, while also providing a theoretical justiﬁcation for the resulting improvement. 1

same-paper 3 0.77624941 138 nips-2009-Learning with Compressible Priors

Author: Volkan Cevher

4 0.67057347 147 nips-2009-Matrix Completion from Noisy Entries

Author: Raghunandan Keshavan, Andrea Montanari, Sewoong Oh

Abstract: Given a matrix M of low-rank, we consider the problem of reconstructing it from noisy observations of a small, random subset of its entries. The problem arises in a variety of applications, from collaborative ﬁltering (the ‘Netﬂix problem’) to structure-from-motion and positioning. We study a low complexity algorithm introduced in [1], based on a combination of spectral techniques and manifold optimization, that we call here O PT S PACE. We prove performance guarantees that are order-optimal in a number of circumstances. 1

5 0.56562436 243 nips-2009-The Ordered Residual Kernel for Robust Motion Subspace Clustering

Author: Tat-jun Chin, Hanzi Wang, David Suter

Abstract: We present a novel and highly effective approach for multi-body motion segmentation. Drawing inspiration from robust statistical model ﬁtting, we estimate putative subspace hypotheses from the data. However, instead of ranking them we encapsulate the hypotheses in a novel Mercer kernel which elicits the potential of two point trajectories to have emerged from the same subspace. The kernel permits the application of well-established statistical learning methods for effective outlier rejection, automatic recovery of the number of motions and accurate segmentation of the point trajectories. The method operates well under severe outliers arising from spurious trajectories or mistracks. Detailed experiments on a recent benchmark dataset (Hopkins 155) show that our method is superior to other stateof-the-art approaches in terms of recovering the number of motions, segmentation accuracy, robustness against gross outliers and computational efﬁciency. 1 Introduction1 Multi-body motion segmentation concerns the separation of motions arising from multiple moving objects in a video sequence. The input data is usually a set of points on the surface of the objects which are tracked throughout the video sequence. Motion segmentation can serve as a useful preprocessing step for many computer vision applications. In recent years the case of rigid (i.e. nonarticulated) objects for which the motions could be semi-dependent on each other has received much attention [18, 14, 19, 21, 22, 17]. Under this domain the afﬁne projection model is usually adopted. Such a model implies that the point trajectories from a particular motion lie on a linear subspace of at most four, and trajectories from different motions lie on distinct subspaces. Thus multi-body motion segmentation is reduced to the problem of subspace segmentation or clustering. To realize practical algorithms, motion segmentation approaches should possess four desirable attributes: (1) Accuracy in classifying the point trajectories to the motions they respectively belong to. This is crucial for success in the subsequent vision applications, e.g. object recognition, 3D reconstruction. (2) Robustness against inlier noise (e.g. slight localization error) and gross outliers (e.g. mistracks, spurious trajectories), since getting imperfect data is almost always unavoidable in practical circumstances. (3) Ability to automatically deduce the number of motions in the data. This is pivotal to accomplish fully automated vision applications. (4) Computational efﬁciency. This is integral for the processing of video sequences which are usually large amounts of data. Recent work on multi-body motion segmentation can roughly be divided into algebraic or factorization methods [3, 19, 20], statistical methods [17, 7, 14, 6, 10] and clustering methods [22, 21, 5]. Notable approaches include Generalized PCA (GPCA) [19, 20], an algebraic method based on the idea that one can ﬁt a union of m subspaces with a set of polynomials of degree m. Statistical methods often employ concepts such random hypothesis generation [4, 17], Expectation-Maximization [14, 6] 1 This work was supported by the Australian Research Council (ARC) under the project DP0878801. 1 and geometric model selection [7, 8]. Clustering based methods [22, 21, 5] are also gaining attention due to their effectiveness. They usually include a dimensionality reduction step (e.g. manifold learning [5]) followed by a clustering of the point trajectories (e.g. via spectral clustering in [21]). A recent benchmark [18] indicated that Local Subspace Afﬁnity (LSA) [21] gave the best performance in terms of classiﬁcation accuracy, although their result was subsequently surpassed by [5, 10]. However, we argue that most of the previous approaches do not simultaneously fulﬁl the qualities desirable of motion segmentation algorithms. Most notably, although some of the approaches have the means to estimate the number of motions, they are generally unreliable in this respect and require manual input of this parameter. In fact this prior knowledge was given to all the methods compared in [18]2 . Secondly, most of the methods (e.g. [19, 5]) do not explicitly deal with outliers. They will almost always breakdown when given corrupted data. These deﬁciencies reduce the usefulness of available motion segmentation algorithms in practical circumstances. In this paper we attempt to bridge the gap between experimental performance and practical usability. Our previous work [2] indicates that robust multi-structure model ﬁtting can be achieved effectively with statistical learning. Here we extend this concept to motion subspace clustering. Drawing inspiration from robust statistical model ﬁtting [4], we estimate random hypotheses of motion subspaces in the data. However, instead of ranking these hypotheses we encapsulate them in a novel Mercer kernel. The kernel can function reliably despite overwhelming sampling imbalance, and it permits the application of non-linear dimensionality reduction techniques to effectively identify and reject outlying trajectories. This is then followed by Kernel PCA [11] to maximize the separation between groups and spectral clustering [13] to recover the number of motions and clustering. Experiments on the Hopkins 155 benchmark dataset [18] show that our method is superior to other approaches in terms of the qualities described above, including computational efﬁciency. 1.1 Brief review of afﬁne model multi-body motion segmentation Let {tf p ∈ R2 }f =1,...,F be the set of 2D coordinates of P trajectories tracked across F frames. In p=1,...,P multi-body motion segmentation the tf p ’s correspond to points on the surface of rigid objects which are moving. The goal is to separate the trajectories into groups corresponding to the motion they belong to. In other words, if we arrange the coordinates in the following data matrix   t11 · · · t1P  . .  ∈ R2F ×P , .. .  T= . (1) . . . tF 1 . . . tF P the goal is to ﬁnd the permutation Γ ∈ RP ×P such that the columns of T · Γ are arranged according to the respective motions they belong to. It turns out that under afﬁne projection [1, 16] trajectories from the same motion lie on a distinct subspace in R2F , and each of these motion subspaces is of dimensions 2, 3 or 4. Thus motion segmentation can be accomplished via clustering subspaces in R2F . See [1, 16] for more details. Realistically actual motion sequences might contain trajectories which do not correspond to valid objects or motions. These trajectories behave as outliers in the data and, if not taken into account, can be seriously detrimental to subspace clustering algorithms. 2 The Ordered Residual Kernel (ORK) First, we take a statistical model ﬁtting point of view to motion segmentation. Let {xi }i=1,...,N be the set of N samples on which we want to perform model ﬁtting. We randomly draw p-subsets from the data and use it to ﬁt a hypothesis of the model, where p is the number of parameters that deﬁne the model. In motion segmentation, the xi ’s are the columns of matrix T, and p = 4 since the model is a four-dimensional subspace3 . Assume that M of such random hypotheses are drawn. i i For each data point xi compute its absolute residual set ri = {r1 , . . . , rM } as measured to the M hypotheses. For motion segmentation, the residual is the orthogonal distance to a hypothesis 2 As conﬁrmed through private contact with the authors of [18]. Ideally we should also consider degenerate motions with subspace dimensions 2 or 3, but previous work [18] using RANSAC [4] and our results suggest this is not a pressing issue for the Hopkins 155 dataset. 3 2 i i subspace. We sort the elements in ri to obtain the sorted residual set ˜i = {rλi , . . . , rλi }, where r 1 M i i the permutation {λi , . . . , λi } is obtained such that rλi ≤ · · · ≤ rλi . Deﬁne the following 1 M 1 M ˜ θi := {λi , . . . , λi } 1 M (2) ˜ as the sorted hypothesis set of point xi , i.e. θi depicts the order in which xi becomes the inlier of the M hypotheses as a ﬁctitious inlier threshold is increased from 0 to ∞. We deﬁne the Ordered Residual Kernel (ORK) between two data points as 1 kr (xi1 , xi2 ) := ˜ Z M/h t ˜ ˜ zt · k∩ (θi1 , θi2 ), (3) t=1 M/h where zt = 1 are the harmonic series and Z = t=1 zt is the (M/h)-th harmonic number. t Without lost of generality assume that M is wholly divisible by h. Step size h is used to obtain the Difference of Intersection Kernel (DOIK) 1 ˜1:α t ˜ ˜ ˜1:α ˜1:α ˜1:α k∩ (θi1 , θi2 ) := (|θi1 t ∩ θi2 t | − |θi1 t−1 ∩ θi2 t−1 |) (4) h ˜a:b where αt = t · h and αt−1 = (t − 1) · h. Symbol θi indicates the set formed by the a-th to ˜i . Since the contents of the sorted hypotheses set are merely permutations of the b-th elements of θ {1 . . . M }, i.e. there are no repeating elements, 0 ≤ kr (xi1 , xi2 ) ≤ 1. ˜ (5) Note that kr is independent of the type of model to be ﬁtted, thus it is applicable to generic statistical ˜ model ﬁtting problems. However, we concentrate on motion subspaces in this paper. Let τ be a ﬁctitious inlier threshold. The kernel kr captures the intuition that, if τ is low, two ˜ points arising from the same subspace will have high normalized intersection since they share many common hypotheses which correspond to that subspace. If τ is high, implausible hypotheses ﬁtted on outliers start to dominate and decrease the normalized intersection. Step size h allows us to quantify the rate of change of intersection if τ is increased from 0 to ∞, and since zt is decreasing, kr will evaluate to a high value for two points from the same subspace. In contrast, kr is always low ˜ ˜ for points not from the same subspace or that are outliers. Proof of satisfying Mercer’s condition. Let D be a ﬁxed domain, and P(D) be the power set of D, i.e. the set of all subsets of D. Let S ⊆ P(D), and p, q ∈ S. If µ is a measure on D, then k∩ (p, q) = µ(p ∩ q), (6) called the intersection kernel, is provably a valid Mercer kernel [12]. The DOIK can be rewritten as t ˜ ˜ k∩ (θi1 , θi2 ) = 1 ˜(αt−1 +1):αt ˜(αt−1 +1):αt (|θ ∩ θi2 | h i1 ˜1:(α ) ˜(α +1):αt | + |θ (αt−1 +1):αt ∩ θ 1:(αt−1 ) |). ˜ ˜ +|θi1 t−1 ∩ θi2 t−1 i1 i2 (7) If we let D = {1 . . . M } be the set of all possible hypothesis indices and µ be uniform on D, each term in Eq. (7) is simply an intersection kernel multiplied by |D|/h. Since multiplying a kernel with a positive constant and adding two kernels respectively produce valid Mercer kernels [12], the DOIK and ORK are also valid Mercer kernels.• Parameter h in kr depends on the number of random hypotheses M , i.e. step size h can be set as a ˜ ratio of M . The value of M can be determined based on the size of the p-subset and the size of the data N (e.g. [23, 15]), and thus h is not contingent on knowledge of the true inlier noise scale or threshold. Moreover, our experiments in Sec. 4 show that segmentation performance is relatively insensitive to the settings of h and M . 2.1 Performance under sampling imbalance Methods based on random sampling (e.g. RANSAC [4]) are usually affected by unbalanced datasets. The probability of simultaneously retrieving p inliers from a particular structure is tiny if points 3 from that structure represent only a small minority in the data. In an unbalanced dataset the “pure” p-subsets in the M randomly drawn samples will be dominated by points from the majority structure in the data. This is a pronounced problem in motion sequences, since there is usually a background “object” whose point trajectories form a large majority in the data. In fact, for motion sequences from the Hopkins 155 dataset [18] with typically about 300 points per sequence, M has to be raised to about 20,000 before a pure p-subset from the non-background objects is sampled. However, ORK can function reliably despite serious sampling imbalance. This is because points from the same subspace are roughly equi-distance to the sampled hypotheses in their vicinity, even though these hypotheses might not pass through that subspace. Moreover, since zt in Eq. (3) is decreasing only residuals/hypotheses in the vicinity of a point are heavily weighted in the intersection. Fig. 1(a) illustrates this condition. Results in Sec. 4 show that ORK excelled even with M = 1, 000. (a) Data in R2F . (b) Data in RKHS Fkr . ˜ Figure 1: (a) ORK under sampling imbalance. (b) Data in RKHS induced by ORK. 3 Multi-Body Motion Segmentation using ORK In this section, we describe how ORK is used for multi-body motion segmentation. 3.1 Outlier rejection via non-linear dimensionality reduction Denote by Fkr the Reproducing Kernel Hilbert Space (RKHS) induced by kr . Let matrix A = ˜ ˜ [φ(x1 ) . . . φ(xN )] contain the input data after it is mapped to Fkr . The kernel matrix K = AT A is ˜ computed using the kernel function kr as ˜ Kp,q = φ(xp ), φ(xq ) = kr (xp , xq ), p, q ∈ {1 . . . N }. ˜ (8) Since kr is a valid Mercer kernel, K is guaranteed to be positive semi-deﬁnite [12]. Let K = ˜ Q∆QT be the eigenvalue decomposition (EVD) of K. Then the rank-n Kernel Singular Value Decomposition (Kernel SVD) [12] of A is 1 1 An = [AQn (∆n )− 2 ][(∆n ) 2 ][(Qn )T ] ≡ Un Σn (Vn )T . n n (9) n Via the Matlab notation, Q = Q:,1:n and ∆ = ∆1:n,1:n . The left singular vectors U is an orthonormal basis for the n-dimensional principal subspace of the whole dataset in Fkr . Projecting ˜ the data onto the principal subspace yields 1 1 B = [AQn (∆n )− 2 ]T A = (∆n ) 2 (Qn )T , (10) n×N where B = [b1 . . . bN ] ∈ R is the reduced dimension version of A. Directions of the principal subspace are dominated by inlier points, since kr evaluates to a high value generally for them, but ˜ always to a low value for gross outliers. Moreover the kernel ensures that points from the same subspace are mapped to the same cluster and vice versa. Fig. 1(b) illustrates this condition. Fig. 2(a)(left) shows the ﬁrst frame of sequence “Cars10” from the Hopkins 155 dataset [18] with 100 false trajectories of Brownian motion added to the original data (297 points). The corresponing RKHS norm histogram for n = 3 is displayed in Fig. 2(b). The existence of two distinct modes, 4 15 Outlier mode Bin count Inlier mode 10 5 0 (a) (left) Before and (right) after outlier removal. Blue dots are inliers while red dots are added outliers. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Vector norm in principal subspace 0.18 0.2 (b) Actual norm histogram of “cars10”. Figure 2: Demonstration of outlier rejection on sequence “cars10” from Hopkins 155. corresponding respectively to inliers and outliers, is evident. We exploit this observation for outlier rejection by discarding data with low norms in the principal subspace. The cut-off threshold ψ can be determined by analyzing the shape of the distribution. For instance we can ﬁt a 1D Gaussian Mixture Model (GMM) with two components and set ψ as the point of equal Mahalanobis distance between the two components. However, our experimentation shows that an effective threshold can be obtained by simply setting ψ as the average value of all the norms, i.e. ψ= 1 N N bi . (11) i=1 This method was applied uniformly on all the sequences in our experiments in Sec. 4. Fig. 2(a)(right) shows an actual result of the method on Fig. 2(a)(left). 3.2 Recovering the number of motions and subspace clustering After outlier rejection, we further take advantage of the mapping induced by ORK for recovering the number of motions and subspace clustering. On the remaining data, we perform Kernel PCA [11] to seek the principal components which maximize the variance of the data in the RKHS, as Fig. 1(b) illustrates. Let {yi }i=1,...,N ′ be the N ′ -point subset of the input data that remains after outlier removal, where N ′ < N . Denote by C = [φ(y1 ) . . . φ(yN ′ )] the data matrix after mapping the data ˜ to Fkr , and by symbol C the result of adjusting C with the empirical mean of {φ(y1 ), . . . , φ(yN ′ )}. ˜ ˜ ˜ ˜ The centered kernel matrix K′ = CT C [11] can be obtained as 1 ˜ K′ = ν T K′ ν, ν = [IN ′ − ′ 1N ′ ,N ′ ], (12) N where K′ = CT C is the uncentered kernel matrix, Is and 1s,s are respectively the s × s identity ˜ ˜ matrix and a matrix of ones. If K′ = RΩRT is the EVD of K′ , then we obtain ﬁrst-m kernel m ˜ principal components P of C as the ﬁrst-m left singular vectors of C , i.e. 1 ˜ Pm = CRm (Ωm )− 2 , (13) where Rm = R:,1:m and Ω1:m,1:m ; see Eq. (9). Projecting the data on the principal components yields 1 D = [d1 . . . dN ′ ] = (Ωm ) 2 (Rm )T , (14) ′ where D ∈ Rm×N . The afﬁne subspace span(Pm ) maximizes the spread of the centered data in the RKHS, and the projection D offers an effective representation for clustering. Fig. 3(a) shows the Kernel PCA projection results for m = 3 on the sequence in Fig. 2(a). The number of clusters in D is recovered via spectral clustering. More speciﬁcally we apply the Normalized Cut (Ncut) [13] algorithm. A fully connected graph is ﬁrst derived from the data, where ′ ′ its weighted adjacency matrix W ∈ RN ×N is obtained as Wp,q = exp(− dp − dq 2 /2δ 2 ), (15) and δ is taken as the average nearest neighbour distance in the Euclidean sense among the vectors in D. The Laplacian matrix [13] is then derived from W and eigendecomposed. Under Ncut, 5 0.1 0.05 0 −0.05 −0.1 0.1 −0.15 0.15 0.08 0.1 0.05 0 −0.05 −0.1 0.06 (a) Kernel PCA and Ncut results. (b) W matrix. (c) Final result for “cars10”. Figure 3: Actual results on the motion sequence in Fig. 2(a)(left). the number of clusters is revealed as the number of eigenvalues of the Laplacian that are zero or numerically insigniﬁcant. With this knowledge, a subsequent k-means step is then performed to cluster the points. Fig. 3(b) shows W for the input data in Fig. 2(a)(left) after outlier removal. It can be seen that strong afﬁnity exists between points from the same cluster, thus allowing accurate clustering. Figs. 3(a) and 3(c) illustrate the ﬁnal clustering result for the data in Fig. 2(a)(left). There are several reasons why spectral clustering under our framework is more successful than previous methods. Firstly, we perform an effective outlier rejection step that removes bad trajectories that can potentially mislead the clustering. Secondly, the mapping induced by ORK deliberately separates the trajectories based on their cluster membership. Finally, we perform Kernel PCA to maximize the variance of the data. Effectively this also improves the separation of clusters, thus facilitating an accurate recovery of the number of clusters and also the subsequent segmentation. This distinguishes our work from previous clustering based methods [21, 5] which tend to operate without maximizing the between-class scatter. Results in Sec. 4 validate our claims. 4 Results Henceforth we indicate the proposed method as “ORK”. We leverage on a recently published benchmark on afﬁne model motion segmentation [18] as a basis of comparison. The benchmark was evaluated on the Hopkins 155 dataset4 which contains 155 sequences with tracked point trajectories. A total of 120 sequences have two motions while 35 have three motions. The sequences contain degenerate and non-degenerate motions, independent and partially dependent motions, articulated motions, nonrigid motions etc. In terms of video content three categories exist: Checkerboard sequences, trafﬁc sequences (moving cars, trucks) and articulated motions (moving faces, people). 4.1 Details on benchmarking Four major algorithms were compared in [18]: Generalized PCA (GPCA) [19], Local Subspace Afﬁnity (LSA) [21], Multi-Stage Learning (MSL) [14] and RANSAC [17]. Here we extend the benchmark with newly reported results from Locally Linear Manifold Clustering (LLMC) [5] and Agglomerative Lossy Compression (ALC) [10, 9]. We also compare our method against Kanatani and Matsunaga’s [8] algorithm (henceforth, the “KM” method) in estimating the number of independent motions in the video sequences. Note that KM per se does not perform motion segmentation. For the sake of objective comparisons we use only implementations available publicly5. Following [18], motion segmentation performance is evaluated in terms of the labelling error of the point trajectories, where each point in a sequence has a ground truth label, i.e. number of mislabeled points . (16) classiﬁcation error = total number of points Unlike [18], we also emphasize on the ability of the methods in recovering the number of motions. However, although the methods compared in [18] (except RANSAC) theoretically have the means to 4 Available at http://www.vision.jhu.edu/data/hopkins155/. For MSL and KM, see http://www.suri.cs.okayama-u.ac.jp/e-program-separate.html/. For GPCA, LSA and RANSAC, refer to the url for the Hopkins 155 dataset. 5 6 do so, their estimation of the number of motions is generally unrealiable and the benchmark results in [18] were obtained by revealing the actual number of motions to the algorithms. A similar initialization exists in [5, 10] where the results were obtained by giving LLMC and ALC this knowledge a priori (for LLMC, this was given at least to the variant LLMC 4m during dimensionality reduction [5], where m is the true number of motions). In the following subsections, where variants exist for the compared algorithms we use results from the best performing variant. In the following the number of random hypotheses M and step size h for ORK are ﬁxed at 1000 and 300 respectively, and unlike the others, ORK is not given knowledge of the number of motions. 4.2 Data without gross outliers We apply ORK on the Hopkins 155 dataset. Since ORK uses random sampling we repeat it 100 times for each sequence and average the results. Table 1 depicts the obtained classiﬁcation error among those from previously proposed methods. ORK (column 9) gives comparable results to the other methods for sequences with 2 motions (mean = 7.83%, median = 0.41%). For sequences with 3 motions, ORK (mean = 12.62%, median = 4.75%) outperforms GPCA and RANSAC, but is slightly less accurate than the others. However, bear in mind that unlike the other methods ORK is not given prior knowledge of the true number of motions and has to estimate this independently. Column Method 1 REF 2 GPCA Mean Median 2.03 0.00 4.59 0.38 Mean Median 5.08 2.40 28.66 28.26 3 4 5 6 LSA MSL RANSAC LLMC Sequences with 2 motions 3.45 4.14 5.56 3.62 0.59 0.00 1.18 0.00 Sequences with 3 motions 9.73 8.23 22.94 8.85 2.33 1.76 22.03 3.19 8 ALC 9 ORK 10 ORK∗ 3.03 0.00 7.83 0.41 1.27 0.00 6.26 1.02 12.62 4.75 2.09 0.05 Table 1: Classiﬁcation error (%) on Hopkins 155 sequences. REF represents the reference/control method which operates based on knowledge of ground truth segmentation. Refer to [18] for details. We also separately investigate the accuracy of ORK in estimating the number of motions, and compare it against KM [8] which was proposed for this purpose. Note that such an experiment was not attempted in [18] since approaches compared therein generally do not perform reliably in estimating the number of motions. The results in Table 2 (columns 1–2) show that for sequences with two motions, KM (80.83%) outperforms ORK (67.37%) by ≈ 15 percentage points. However, for sequences with three motions, ORK (49.66%) vastly outperforms KM (14.29%) by more than doubling the percentage points of accuracy. The overall accuracy of KM (65.81%) is slightly better than ORK (63.37%), but this is mostly because sequences with two motions form the majority in the dataset (120 out of 155). This leads us to conclude that ORK is actually the superior method here. Dataset Column Method 2 motions 3 motions Overall Hopkins 155 1 2 KM ORK 80.83% 67.37% 14.29% 49.66% 65.81% 63.37% Hopkins 155 + Outliers 3 4 KM ORK 00.00% 47.58% 100.00% 50.00% 22.58% 48.13% Table 2: Accuracy in determining the number of motions in a sequence. Note that in the experiment with outliers (columns 3–4), KM returns a constant number of 3 motions for all sequences. We re-evaluate the performance of ORK by considering only results on sequences where the number of motions is estimated correctly by ORK (there are about 98 ≡ 63.37% of such cases). The results are tabulated under ORK∗ (column 10) in Table 1. It can be seen that when ORK estimates the number of motions correctly, it is signiﬁcantly more accurate than the other methods. Finally, we compare the speed of the methods in Table 3. ORK was implemented and run in Matlab on a Dual Core Pentium 3.00GHz machine with 4GB of main memory (this is much less powerful 7 than the 8 Core Xeon 3.66GHz with 32GB memory used in [18] for the other methods in Table 3). The results show that ORK is comparable to LSA, much faster than MSL and ALC, but slower than GPCA and RANSAC. Timing results of LLMC are not available in the literature. Method 2 motions 3 motions GPCA 324ms 738ms LSA 7.584s 15.956s MSL 11h 4m 1d 23h RANSAC 175ms 258ms ALC 10m 32s 10m 32s ORK 4.249s 8.479s Table 3: Average computation time on Hopkins 155 sequences. 4.3 Data with gross outliers We next examine the ability of the proposed method in dealing with gross outliers in motion data. For each sequence in Hopkins 155, we add 100 gross outliers by creating trajectories corresponding to mistracks or spuriously occuring points. These are created by randomly initializing 100 locations in the ﬁrst frame and allowing them to drift throughout the sequence according to Brownian motion. The corrupted sequences are then subjected to the algorithms for motion segmentation. Since only ORK is capable of rejecting outliers, the classiﬁcation error of Eq. (16) is evaluated on the inlier points only. The results in Table 4 illustrate that ORK (column 4) is the most accurate method by a large margin. Despite being given the true number of motions a priori, GPCA, LSA and RANSAC are unable to provide satisfactory segmentation results. Column Method Mean Median Mean Median 1 2 3 4 GPCA LSA RANSAC ORK Sequences with 2 motions 28.66 24.25 30.64 16.50 30.96 26.51 32.36 10.54 Sequences with 3 motions 40.61 30.94 42.24 19.99 41.30 27.68 43.43 8.49 5 ORK∗ 1.62 0.00 2.68 0.09 Table 4: Classiﬁcation error (%) on Hopkins 155 sequences with 100 gross outliers per sequence. In terms of estimating the number of motions, as shown in column 4 in Table 2 the overall accuracy of ORK is reduced to 48.13%. This is contributed mainly by the deterioration in accuracy on sequences with two motions (47.58%), although the accuracy on sequences with three motions are maintained (50.00%). This is not a surprising result, since sequences with three motions generally have more (inlying) point trajectories than sequences with two motions, thus the outlier rates for sequences with three motions are lower (recall that a ﬁxed number of 100 false trajectories are added). On the other hand, the KM method (column 3) is completely overwhelmed by the outliers— for all the sequences with outliers it returned a constant “3” as the number of motions. We again re-evaluate ORK by considering results from sequences (now with gross outliers) where the number of motions is correctly estimated (there are about 75 ≡ 48.13% of such cases). The results tabulated under ORK∗ (column 5) in Table 4 show that the proposed method can accurately segment the point trajectories without being inﬂuenced by the gross outliers. 5 Conclusions In this paper we propose a novel and highly effective approach for multi-body motion segmentation. Our idea is based on encapsulating random hypotheses in a novel Mercel kernel and statistical learning. We evaluated our method on the Hopkins 155 dataset with results showing that the idea is superior other state-of-the-art approaches. It is by far the most accurate in terms of estimating the number of motions, and it excels in segmentation accuracy despite lacking prior knowledge of the number of motions. The proposed idea is also highly robust towards outliers in the input data. Acknowledgements. We are grateful to the authors of [18] especially Ren´ Vidal for discussions e and insights which have been immensely helpful. 8 References [1] T. Boult and L. Brown. Factorization-based segmentation of motions. In IEEE Workshop on Motion Understanding, 1991. [2] T.-J. Chin, H. Wang, and D. Suter. Robust ﬁtting of multiple structures: The statistical learning approach. In ICCV, 2009. [3] J. Costeira and T. Kanade. A multibody factorization method for independently moving objects. IJCV, 29(3):159–179, 1998. [4] M. A. Fischler and R. C. Bolles. Random sample concensus: A paradigm for model ﬁtting with applications to image analysis and automated cartography. Comm. of the ACM, 24:381–395, 1981. [5] A. Goh and R. Vidal. Segmenting motions of different types by unsupervised manifold clustering. In CVPR, 2007. [6] A. Gruber and Y. Weiss. Multibody factorization with uncertainty and missing data using the EM algorithm. In CVPR, 2004. [7] K. Kanatani. Motion segmentation by subspace separation and model selection. In ICCV, 2001. [8] K. Kanatani and C. Matsunaga. Estimating the number of independent motions for multibody segmentation. In ACCV, 2002. [9] Y. Ma, H. Derksen, W. Hong, and J. Wright. Segmentation of multivariate mixed data via lossy coding and compression. TPAMI, 29(9):1546–1562, 2007. [10] S. Rao, R. Tron, Y. Ma, and R. Vidal. Motion segmentation via robust subspace separation in the presence of outlying, incomplete, or corrupted trajectories. In CVPR, 2008. [11] B. Sch¨ lkopf, A. Smola, and K. R. M¨ ller. Nonlinear component analysis as a kernel eigeno u value problem. Neural Computation, 10:1299–1319, 1998. [12] J. Shawe-Taylor and N. Cristianini. Kernel methods for pattern analysis. Cambridge University Press, 2004. [13] J. Shi and J. Malik. Normalized cuts and image segmentation. TPAMI, 22(8):888–905, 2000. [14] Y. Sugaya and K. Kanatani. Geometric structure of degeneracy for multi-body motion segmentation. In Workshop on Statistical Methods in Video Processing, 2004. [15] R. Toldo and A. Fusiello. Robust multiple structures estimation with J-Linkage. In ECCV, 2008. [16] C. Tomasi and T. Kanade. Shape and motion from image streams under orthography. IJCV, 9(2):137–154, 1992. [17] P. Torr. Geometric motion segmentation and model selection. Phil. Trans. Royal Society of London, 356(1740):1321–1340, 1998. [18] R. Tron and R. Vidal. A benchmark for the comparison of 3-D motion segmentation algorithms. In CVPR, 2007. [19] R. Vidal and R. Hartley. Motion segmentation with missing data by PowerFactorization and Generalized PCA. In CVPR, 2004. [20] R. Vidal, Y. Ma, and S. Sastry. Generalized Principal Component Analysis (GPCA). TPAMI, 27(12):1–15, 2005. [21] J. Yan and M. Pollefeys. A general framework for motion segmentation: independent, articulated, rigid, non-rigid, degenerate and non-degenerate. In ECCV, 2006. [22] L. Zelnik-Manor and M. Irani. Degeneracies, dependencies and their implications on multibody and multi-sequence factorization. In CVPR, 2003. [23] W. Zhang and J. Koseck´ . Nonparametric estimation of multiple structures with outliers. In a Dynamical Vision, ICCV 2005 and ECCV 2006 Workshops, 2006. 9

6 0.5335902 104 nips-2009-Group Sparse Coding

7 0.52511525 158 nips-2009-Multi-Label Prediction via Sparse Infinite CCA

8 0.52454591 254 nips-2009-Variational Gaussian-process factor analysis for modeling spatio-temporal data

9 0.52336597 3 nips-2009-AUC optimization and the two-sample problem

10 0.5226379 169 nips-2009-Nonlinear Learning using Local Coordinate Coding

11 0.52245593 137 nips-2009-Learning transport operators for image manifolds

12 0.52061731 100 nips-2009-Gaussian process regression with Student-t likelihood

13 0.51983368 167 nips-2009-Non-Parametric Bayesian Dictionary Learning for Sparse Image Representations

14 0.51950598 191 nips-2009-Positive Semidefinite Metric Learning with Boosting

15 0.51930666 70 nips-2009-Discriminative Network Models of Schizophrenia

16 0.51888335 30 nips-2009-An Integer Projected Fixed Point Method for Graph Matching and MAP Inference

17 0.51631081 17 nips-2009-A Sparse Non-Parametric Approach for Single Channel Separation of Known Sounds

18 0.51587337 61 nips-2009-Convex Relaxation of Mixture Regression with Efficient Algorithms

19 0.51569617 145 nips-2009-Manifold Embeddings for Model-Based Reinforcement Learning under Partial Observability

20 0.51554173 35 nips-2009-Approximating MAP by Compensating for Structural Relaxations