A systematic review of structured sparse learning
- 290 Downloads
- 2 Citations
Abstract
High dimensional data arising from diverse scientific research fields and industrial development have led to increased interest in sparse learning due to model parsimony and computational advantage. With the assumption of sparsity, many computational problems can be handled efficiently in practice. Structured sparse learning encodes the structural information of the variables and has been quite successful in numerous research fields. With various types of structures discovered, sorts of structured regularizations have been proposed. These regularizations have greatly improved the efficacy of sparse learning algorithms through the use of specific structural information. In this article, we present a systematic review of structured sparse learning including ideas, formulations, algorithms, and applications. We present these algorithms in the unified framework of minimizing the sum of loss and penalty functions, summarize publicly accessible software implementations, and compare the computational complexity of typical optimization methods to solve structured sparse learning problems. In experiments, we present applications in unsupervised learning, for structured signal recovery and hierarchical image reconstruction, and in supervised learning in the context of a novel graph-guided logistic regression.
Key words
Sparse learning Structured sparse learning Structured regularizationCLC number
TP3911 Introduction
Rapid improvement in sensing technologies has created high dimensional data in research fields and industries with many features and a huge number of samples. The increasing data volume becomes a great challenge for contemporary statistical learning algorithms (John Lu, 2010) in various research areas, including high-resolution imaging (Bruckstein et al., 2009), target tracking (Zhang et al., 2012; 2013; 2014; 2015a; 2015b), astronomical data processing (Borne, 2009), genomics (Kim and Xing, 2014), functional and longitudinal processing (Jenatton et al., 2012), and warehouse data analysis in business (Fan et al., 2011). For example, astronomical projects produce more than 10^{9} pixels every 20 s and terabytes of data in a single evening (Borne, 2009). Financial data is measured with hundreds of financial instruments and tracked over time with 10^{6} trades per second in high-frequency trading (Fan et al., 2011).
It is almost impossible to learn a consistent model with high accuracy, model explicability, and computational efficiency at the same time, unless one assumes that the sample size is much larger than the feature size (Candès and Tao, 2007). However, in high-dimensional settings, the dimension of the feature is often the same as, or even larger than the sample size. Therefore, traditional methods face significant challenges, ranging from theoretical analysis, efficient algorithm design, to model estimation and interpretation. Note that consistent estimators may be obtained if additional assumptions are imposed on the traditional models (Negahban et al., 2012). A widely used constraint is that models should be sparse for high-dimensional problems (Bach et al., 2011).
1.1 Sparsity
In a sparse model, only a small number of variables are non-zero among all the variables in the model. The sparsity assumption is typically associated with desired interesting properties such as succinct interpretation, fast evaluation of the model, statistical robustness (sparsity is usually associated with robust statistical performance), and computational advantages, which appeals to a great many researchers.
Sparsity is preferred in learning problems with high-dimensional data. In many applications, though the raw data is high dimensional, the intrinsic dimension is relatively low. For example, in bioinformatics, different high-dimensional genes may belong to the same functional group; in multi-task learning, several estimators are expected to share common types of covariates. In fact, it is now common sense that sparsity is a powerful assumption for contemporary machine learning algorithms (Bach et al., 2012a).
Sparsity is considered to be one of the most significant philosophical and aesthetic principles that have ever existed. It is also known as Occam’s razor (Rasmussen and Ghahramani, 2001), “Entities should not be multiplied without necessity”, by William Ockham in the 13th century. The parsimony principle has been addressed again and again, which has led to several beautiful results, such as minimal description length (MDL) (John Lu, 2010). Modern sparse learning methods were introduced by Wrinch and Jeffreys (1921), who expressed the sparsity of models in physics as the non-zero number of learning variables. This concept is very close to today’s definition of sparsity. Since then, numerous tools (see Mairal et al. (2014) and multiple references therein) have been developed in the statistics community to build sparsity-related models, which have greatly improved the explicability of the models, and dramatically decreased the computational cost of the model in the prediction procedure. With the efforts of researchers and engineers, sparse learning has become a popular tool with the development of theoretical frameworks and various efficient algorithms. The theoretical frameworks range from the original idea in underdetermined linear systems (see Bruckstein et al. (2009) and references therein), signal processing (Chen and Donoho, 1994) and statistical learning (Tibshirani, 1996), the computational complexity of the ℓ_{0}-norm regularized problem, the uniqueness of the solution for the ℓ_{0}-norm regularized optimization (see Donoho and Huo (2001), Tropp et al. (2003), Elad (2010), and references therein), the model selection consistency of the convex relaxation from ℓ_{0}-norm to ℓ_{1}-norm (Candès et al., 2006; Zhao and Yu, 2006; Candès and Tao, 2007; Candès, 2008; Candès and Recht, 2009; Zhang, 2009), to the statistical analysis of extended algorithms (Elad, 2010; Jenatton, 2011). Proposed algorithms range from greedy algorithms for ℓ_{0}-norm regularized methods (Tropp, 2004) to convex optimization methods after convex approximation (see Friedman et al. (2007), Beck and Teboulle (2009), Jenatton et al. (2011), Yang and Yuan (2013), and references therein).
Examples of nonconvex penalty r(x)
Name | r(x) |
---|---|
LSP | γ log(l + ∣x∣/θ)(θ > 0) |
SCAD | \(\gamma \int\nolimits_0^{\vert x\vert} {\min \left( {1,\;{{{{[\theta \gamma - y]}_ +}} \over {(\theta - 1)\gamma}}} \right)\;{\rm{d}}y\;\;(\theta > 2)}\) |
MCP | \(\gamma \int\nolimits_0^{\vert x\vert} {{{\left[ {1 - {y \over {\theta \gamma}}} \right]}_ +}\;{\rm{d}}y = \left\{ {\begin{array}{*{20}c}{\gamma \vert x\vert - {{{x^2}} \over {2\theta}},} & {\vert x\vert \leq \theta \gamma ,} \\ {\theta {\gamma ^2}/2,\quad \quad} & {\vert x\vert \leq \theta \gamma ,} \\ {} & {(\theta > 0)} \end{array}} \right.}\) |
Capped-ℓ_{1} | γ min (∣x∣, 0) (θ > 0) |
1.2 Structured sparsity
Structured sparse learning is commonly used in two situations. First, structured sparse learning is used given the prior knowledge that the model should be structured sparsely. Second, to make the model more interpretable or easier to use in the following procedures, even if the underlying problem does not admit assumed structured sparse solutions, one looks for the best structured sparse approximation.
1.3 Aim and scope of this paper
Structured sparsity has been widely used in practical problems, including model-based compressive sensing (Baraniuk et al., 2010; Asaei et al., 2011a; Duarte and Eldar, 2011; Chen et al., 2014), signal processing (Bach and Jordan, 2006; Asaei et al., 2011a; 2014a; 2014b; Najafian, 2016), computer vision (Jenatton et al., 2009; Kim et al., 2013; Chen and Huang, 2014; Karygianni and Frossard, 2014; Xiao et al., 2016), bioinformatics (Wille and Bühlmann, 2006; Zhang SZ et al., 2011; Kim and Xing, 2012), and recommendation systems (Koren et al., 2009; Takacs et al., 2009; Rendle and Schmidt-Thieme, 2010; Zhang ZK et al., 2011). This article focuses mainly on their formulations and algorithms. The models and the algorithms are relatively independent (Xu et al., 2005; Zhang et al., 2006; Xu et al., 2007; Yuan et al., 2015; Hu and Yu, 2016; Xie et al., 2016; Xie and Tong, 2016; Zhu et al., 2016), so is in the privacy research field (Sun et al., 2015a; 2015b; Wu et al., 2016).
Note that there are other books and articles that offer diverse perspectives on sparse learning methods, including Elad (2010) and Mallat (2008) from a perspective of signal processing, and Bach et al. (2012b) in a view of optimization. However, this article focuses on intuitive formulations, their variants, and the algorithms to solve them.
1.4 Notations
Vectors are denoted by bold lower-case letters and matrices by upper-case ones. ∥x∥_{0} is the number of non-zero elements in a vector x, ∥x∥_{1} is the sum of absolute values of elements in a vector x, the ℓ_{ q }-norm of a vector x ∈ ℝ^{ n } is defined as \({\left\Vert x \right\Vert _q}\;\;: = {({\sum\nolimits_{i = 1}^n {\vert {x_i}\vert} ^q})^{1/q}}\) for q > 0, and ∥x∥_{∞} ≔ max_{i=1, 2,…, n} ∣x_{ i }∣, where x_{ i } denotes the ith coordinate of x. The Frobenius norm of a matrix X ∈ ℝ^{d×n} is defined as \({\left\Vert X \right\Vert _{\rm{F}}}\;\;: = {(\sum\nolimits_{i = 1}^n {\sum\nolimits_{j = 1}^p {x_{ij}^2}} )^{1/2}}\), where x_{ ij } denotes the entry of X at the ith row and jth column. X_{ i } denotes the ith column of X.
2 Comparisons of different structured sparsities and computational complexity of optimization methods
In this section, we summarize all these structured sparsities presented in this article and compare the computational complexity of typical optimization methods of these structured sparse learning problems.
In Table 2, we list formulations of these structured sparse learning problems, their corresponding optimization algorithms, and public available software implementations.
An overview of formulations, processing algorithms, and software packages for Lasso and its structured extensions
Method | Loss, l(x) | Regularization, R(x) | Optimization algorithms | Software packages |
---|---|---|---|---|
Lasso (Tibshirani, 1996) | \(\Vert y - Ax\Vert _{\rm{F}}^2\) | λ∥x∥_{1} | Generic QP methods after reformulation (Bach et al., 2012a), alternating direction methods (Boyd et al., 2011), proximal methods (Parikh and Boyd, 2014), block coordinate descent methods (Tseng and Yun, 2009; Wen et al., 2012; Peng et al., 2016), iteratively reweighted methods (Chartrand and Yin, 2008; Lai et al., 2013), working-set and homotopy methods (Bach et al., 2012a) | CVX (Grant and Boyd, 2013), SDPT3 (Toh et al., 2006), YALL1 (Zhang Y et al., 2011), SPGL1 (van den Berg and Friedlander, 2007), SLEP (Liu J et al., 2009), SPAMs (Mairal et al., 2011), SparseLab (Donoho et al., 2007) |
Grouped Lasso (Yuan and Lin, 2006) | \(\Vert y - {\beta_0} - \sum\limits_{i = 1}^g {X_i^{\rm{T}}{\beta_i}} \Vert _{\rm{F}}^2\) | \(\lambda \sum\nolimits_{i = 1}^g {\sqrt {{p_i}} \Vert {\beta_i}\vert {\vert _2}}\) | Generic SOCP methods after reformulation (Bach, 2008b), alternating direction methods (Boyd et al., 2011), proximal methods (Parikh and Boyd, 2014), block coordinate descent methods (Tseng and Yun, 2009; Wen et al., 2012; Peng et al., 2016), iteratively reweighted methods (Chartrand and Yin, 2008; Lai et al., 2013), working-set and homotopy methods (Bach et al., 2012a) | CVX (Grant and Boyd, 2013), SDPT3 (Toh et al., 2006), YALL1 (Zhang Y et al., 2011), SLEP (Liu J et al., 2009), SPAMs (Mairal et al., 2011) |
Fused Lasso (Tibshirani et al., 2005) | \(\sum\limits_{i = 1}^N {{{({y_i} - {\beta_0} - \sum\limits_{j = 1}^p {{x_{ij}}{\beta_j}} )}^2}}\) | \(\begin{array}{*{20}c}{{\lambda_1}{{\sum\nolimits_{j = 1}^p {\Vert {\beta_j}\Vert}}_1} + \quad \quad \quad} \\ {{\lambda_2}{{\sum\nolimits_{j = 1}^{p - 1} {\Vert {\beta_{j + 1}} - {\beta_j}\Vert}}_1}} \end{array}\) | Generic SDP methods after reformulation (Selesnick and Bayram, 2014), alternating direction methods (Boyd et al., 2011), proximal methods (Parikh and Boyd, 2014), block coordinate descent methods (Tseng and Yun, 2009; Wen et al., 2012; Peng et al., 2016), iteratively reweighted methods (Chartrand and Yin, 2008; Lai et al., 2013), working-set and homotopy methods (Bach et al., 2012a) | CVX (Grant and Boyd, 2013), SDPT3 (Toh et al., 2006), SLEP (Liu J et al., 2009), SPAMs (Mairal et al., 2011) |
Hierarchical Lasso (Zhao et al., 2009) | \(\Vert y - {\beta_0} - \sum\limits_{i = 1}^{g}X_{i}^{\rm{T}}\beta_{i} \Vert _{\rm{F}}^2\) | \(\lambda \sum\nolimits_{g \in {\mathcal G}} {{w_g}\left\Vert {{\alpha_g}} \right\Vert}\) | Generic SDP methods after reformulation (Bach, 2008a), alternating direction methods (Boyd et al., 2011), proximal methods (Parikh and Boyd, 2014), block coordinate descent methods (Tseng and Yun, 2009; Wen et al., 2012; Peng et al., 2016), iteratively reweighted methods (Chartrand and Yin, 2008; Lai et al., 2013), working-set and homotopy methods (Bach et al., 2012a) | CVX (Grant and Boyd, 2013), SDPT3 (Toh et al., 2006), SLEP (Liu J et al., 2009), SPAMs (Mairal et al., 2011) |
Graphical Lasso (Meinshausen and Bühlmann, 2006) | log det Θ − trace(SΘ) | λ∥Θ∥_{1} | Generic SDP methods after reformulation (Bach, 2008a), alternating direction methods (Boyd et al., 2011), proximal methods (Parikh and Boyd, 2014), block coordinate descent methods (Tseng and Yun, 2009; Wen et al., 2012; Peng et al., 2016), iteratively reweighted methods (Chartrand and Yin, 2008; Lai et al., 2013), working-set and homotopy methods (Bach et al., 2012a) | CVX (Grant and Boyd, 2013), SDPT3 (Toh et al., 2006), SLEP (Liu J et al., 2009), SPAMs (Mairal et al., 2011) |
An overview of optimization methods’ computational complexity
Algorithm | Formulation | Convex | Strongly convex | Notes |
---|---|---|---|---|
Subgradient descent | \(\min\limits_{x \in \mathbb{X}^{n}} \;f(x)\) | O(1/ε^{2}) (Nesterov, 2004) | O(1/ε) (Lacoste-Julien et al., 2012) | Cannot be improved upon without further assumptions |
Mirror descent | \(\min\limits_{x \in {\mathcal C}} \;f(x)\) | O(1/ε^{2}) (Beck and Teboulle, 2003) | O(1/ε) (Nemirovski, 2004) | Different parameterizations result in gradient descent and exponentiated gradient descent |
Dual averaging | \(\min\limits_{x \in {\mathcal C}} \;f(x)\) | O(1/ε^{2}) (Nesterov, 2009) | O(log(1/ε)) (Suzuki, 2013) | Cannot be improved upon without further assumptions |
Gradient descent | \(\min\limits_{x \in \mathbb{X}^{n}} \;f(x)\) | O(1/ε) (Nesterov, 2004) | O(log(1/ε)) (Hazan et al., 2007) | Applicable when f(x) is a strongly convex function |
Accelerated gradient descent | \(\min\limits_{x \in\mathbb{X}^{n}} \;f(x)\) | \(O(1/\sqrt \varepsilon )\) (Su et al., 2014) | O(log(1/ε)) (Tseng, 2008) | Applicable when f(x) is differentiable. Cannot be improved upon without further assumptions. Has better constants than gradient descent for the strongly convex case |
Proximal gradient descent | \(\min\limits_{x \in {\mathcal C}} \;f(x) + g(x)\) | O(1/ε) (Combettes and Pesquet, 2011) | O(log(1/ε)) (Suzuki, 2013) | Applicable when f(x) is differentiable and prox_{ τtg }(x) is easily computable |
Accelerated proximal gradient descent | \(\min\limits_{x \in {\mathcal C}} \;f(x) + g(x)\) | \(O(1/\sqrt \varepsilon )\) (Mairal, 2013) | O(log(1/ε)) (Lin et al., 2015) | Applicable when f(x) is differentiable and prox_{ τtg }(x) is easily computable. Has better constants than proximal gradient descent for the strongly convex case |
Frank-Wolfe algorithm/conditional gradient algorithm | \(\min\limits_{x \in {\mathcal C}} \;f(x)\) | O(1/ε) (Jaggi, 2013) | \(O(1/\sqrt \varepsilon )\) (Garber and Hazan, 2015) | Applicable when \({\mathcal C}\) is bounded and \({h_g}(x)\;\; =\) \(\arg {\min\nolimits_{x \in {\mathcal C}}}\) < g, x > is easily computable. Most useful when\({\mathcal C}\) is a polytope in a very high dimensional space with sparse extrema |
3 Grouped structured sparsity
In many regression problems, the variables are predefined in groups as prior knowledge in practical situations such as the analysis-of-variance (ANOVA). In this setting, the typical goal is to choose major effects and interactions among variables. For example, in supervised learning problems, to generate the classification label, variables are organized as collections of categorical predictors, and the selection of significant variables corresponds to the selection of groups of variables. To handle these problems, grouped structured sparsity has been proposed, and these extended methods are often called ‘grouped Lasso’ (Yuan and Lin, 2006).
3.1 Formulations of grouped structured sparsity
The grouped Lasso was studied and generalized by Yuan and Lin (2006). Assume that the number of groups is fixed and finite, and predictors are divided into g groups with p_{ i } as the number of predictors in the ith group G_{ i }, \(p = \sum\nolimits_{i = 1}^g {{p_i}}\), and G_{ i } ∩ G_{ j } = ⊘.
The sparsity of the solution depends on the magnitude of the tuning parameter λ, and exploits the non-differentiability of ∥β_{ i }∥_{2} at β_{ i } = 0. Note that the grouped Lasso estimates and the grouped sparsity pattern converges to the correct patterns in probability (Bach, 2008b).
3.2 Algorithms for grouped structured sparsity
The grouped Lasso optimization problem (3) can be solved through a coordinate gradient descent algorithm, which is applicable to a broad class of convex loss functions, and convergence of the algorithm is established (see Hong et al. (2015) and references therein). The blockwise coordinate descent (BCD) algorithm was also used to solve the problem in Meier et al. (2008) and Liu H et al. (2009). However, the BCD method may get stuck in ill conditions, and in Vincent and Hansen (2014), a modified BCD algorithm was proposed, which first computes a descent direction at the given point, and then uses a line search to find the next starting point. There are also publicly accessible software implementations that can be used to solve the problem. For example, the SLEP package (Liu J et al., 2009) was used in Xie and Xu (2014). More available implementations are listed in Table 2.
The performance of these algorithms has been comparatively studied in Rakotomamonjy (2011), who concluded that depending on the performance measure, greedy approaches and iterative re-weighted algorithms are more efficient in either computational complexity or sparsity recovered.
3.3 Applications of grouped structured sparsity
There are numerous applications to pursue grouped sparsity. In computer vision, each group corresponds to different data sources or data types, and different data sources could be referred to as views. In speech and signal processing, similar groups represent different frequency bands (McAuley et al., 2005). In Vincent and Hansen (2014), grouped Lasso was used in multinomial classification and solved with a coordinate gradient descent method. In document processing, Bengio et al. (2009) combined grouped sparse learning with a bag-of-words document representation.
4 Fused structured sparsity
In many real-world applications, coefficients are organized in a specific order and have local constancy. For instance, organizing variables in blocks as prior knowledge may produce a better result. In this setting, it is necessary to extend Lasso to exploit the ordered structure, and the extension is called ‘fussed Lasso’ (Tibshirani et al., 2005).
4.1 Formulations of fused structured sparsity
A generalized two-dimensional fused Lasso was proposed in Friedman et al. (2007). In this model, parameters are laid out in a grid of pixels with a 2D total variation (TV) norm (Rudin et al., 1992), which is usually used in image denoising, image smoothing, and data recovery. The general strategy for two-dimensional fused Lasso can be directly applied in higher-order problems, formulated as a tensor with a higher-order TV-norm as regularization.
4.2 Algorithms to solve fused structured sparsity
The fused Lasso is a strictly convex problem in β. For 1D fused Lasso, SQOPT (Gill et al., 2008) can be used directly, in which there is a two-phase, active-set algorithm designed for quadratic programming problems with sparse linear constraints. For a class of convex optimization problems, a coordinate descent algorithm was presented in Friedman et al. (2007), which is a one-at-a-time coordinate-wise descent algorithm, and can be generalized to solve the 2D fused Lasso or even higher dimensional cases.
To handle large-scale fused Lasso problems, Ye and Xie (2011) proposed an iterative algorithm based on the split Bregman method to solve a class of large-scale fused Lasso problems. Wang LC et al. (2013) presented an augmented Lagrangian method (ALM) for general convex loss. Li et al. (2014) proposed a fast linearized alternating direction method to solve the general Lasso model, and Qiao et al. (2016b; 2016c) improved the method to solve the structed nonconvex problems.
Experiment results in Hoefling (2010) showed that Tibshirani’s and Friedman’s algorithms are clear and fast, and are state-of-the-art methods. The linearized method in Li et al. (2014) can be used to handle larger-scale problems.
4.3 Applications of fused structured sparsity
Fused Lasso is validated in protein mass spectroscopy (MS), gene expression, and image smoothing problems. The protein MS, which holds great promise for biomarker identification and proteomics profiling, was also used as a motivating example to demonstrate the efficacy of fused Lasso in Tibshirani et al. (2005) and Tibshirani and Wang (2008). As demonstrated in Huang et al. (2005), another important application of fused Lasso is the reconstruction of copy numbers from comparative genomic hybridization (CGH) data arrays. For images with smoothness among pixels, fused Lasso can achieve a rather good performance (Friedman et al., 2007).
5 Hierarchically structured sparsity
In hierarchical sparsity, the variables are organized hierarchically (Xu et al., 2011) or integrated into a tree, and form a union of potentially overlapping groups that were defined previously. The hierarchical sparsity may be achieved through various extended sparsity-inducing norms, and the extension is often called ‘hierarchical Lasso’ (Zhao et al., 2009).
5.1 Formulations of hierarchically structured sparsity
Hierarchical Lasso assumes that the p variables are assigned to the nodes of a tree T, or a forest. In this setting, if a feature is selected then it implies that all its ancestors in T have already been selected, and if a node is not selected, then its descendants are not selected.
In compressive sensing, Baraniuk et al. (2010) presented tree sparsity in the context of sparse wavelet decompositions. The consistency of the sparse estimator of potentially overlapping groups was given in Jacob et al. (2009). In sparse coding, Jenatton et al. (2010) proposed an extension that the atoms are further assumed to be embedded in a tree, which was achieved using tree-structured sparse regularization norms.
5.2 Algorithms of hierarchically structured sparsity
The BLasso algorithm (Zhao and Yu, 2007), which was derived from a coordinate descent method with a fixed step size applied to the general Lasso loss function, can be used to solve the minimization problem. Zhao et al. (2009) extended BLasso and proposed the hiCAP algorithm for hierarchical variables selection, which was valid for the ℓ_{2}-loss when γ_{0} = 1, γ_{ k } = ∞, or a tree-structured hierarchy in graph representation. Considering the formulation’s nonseparability and non-smoothness, Chen et al. (2012) proposed the smoothing proximal gradient (SPG) method, which combines a smoothing technique with an effective proximal gradient method to solve structured sparse regression problems with a smooth convex loss.
Proximal methods (Parikh and Boyd, 2014), which extend the projection operator to a convex set, have recently been shown effective in solving variational problems. To accelerate the convergence, Mosci et al. (2010) added a strictly convex function to the objective function and the experiment results showed that it reduces the number of substantial optimization iterations. After introducing auxiliary variables, Micchelli et al. (2013) used an alternating minimization algorithm with a projection procedure to solve the problem and established the theorem of convergence. Combined with an active set strategy, Villa et al. (2014) accelerated the proximal method by using a new active set strategy to compute the proximal operator. When the objective function is strongly convex, Xiao and Zhang (2014) proposed a proximal stochastic gradient method iterated in an incremental way, which provides an efficient way to solve large-scale problems.
5.3 Applications of hierarchically structured sparsity
In wavelet decompositions, it is natural to organize them in a tree because of their multi-scale structure, and it benefits image compression and de-noising (Baraniuk et al., 2010; Huang et al., 2011). In dictionary learning, Jenatton (2011) used hierarchical dictionary learning in image restoration and performed multi-scale mining of fMRI data for the prediction of simple cognitive tasks (Jenatton et al., 2012). In genetics, Kim and Xing (2010) used it to exploit the tree structure of gene networks for multi-task regression. In topic models, Blei et al. (2010) proposed a hierarchical model of latent variables based on Bayesian non-parametric methods to model hierarchies of topics.
6 Graphically structured sparsity
The graph is a powerful data structure in model construction for a lot of machine learning algorithms, such as the graphical model (Wainwright and Jordan, 2008) and high-dimensional model selection (Meinshausen and Bühlmann, 2006). Sparse graphs have a relatively small number of edges, and are economical to use with good interpretability. The problem of estimating sparse graphs may be resolved by ‘graphical Lasso’ (Banerjee et al., 2008), which is an extension of Lasso on the inverse covariance matrix.
6.1 Formulations of graphically structured sparsity
A graph G consists of a set of vertices N, and an edge set E. In undirected graphical models, each vertex represents a random variable, and each edge represents the dependent relationship between the two vertices. The absence of an edge between two vertices has a special meaning: the corresponding random variables are conditionally independent, given that the rest of the variables are known.
Wermuth (1976) showed that if graph G is Gaussian distributed, then it has the property that conditional independence of vertices corresponds to non-zero entries in the precision matrix. Model selection for undirected Gaussian graphical models is equivalent to selecting non-zero elements in the precision matrix. Dempster (1972) named the problem ‘covariance selection’. Considering that the values of the precision matrix are part of continuous variables, it is natural to extend variable selection to edge selection on the graph (see Banerjee et al. (2008) and references therein).
6.2 Algorithms of graphically structured sparsity
Following Banerjee’s work, Friedman et al. (2008) proposed an algorithm named ‘graphical Lasso’ using a coordinate descent method. Motivated by the success of convex relaxation for the rank-minimization problem, Chandrasekaran et al. (2012) introduced a regularized maximum normal likelihood decomposition framework with a trace norm penalty term, and Ma et al. (2013) developed a proximal gradient based alternating direction method of multipliers to solve these problems.
In practice, tuning the parameter is essential for the results. Yuan and Lin (2007) proposed a Bayesian information criterion (BIC) type criterion for the selection of the tuning parameter.
6.3 Applications of graphically structured sparsity
Graphical Lasso has been applied to various research fields, including gene network discovery and social-network data analysis.
Jones and West (2005) applied graphical Lasso to the analysis of gene expression data, which consists of 8408 variables and has roughly a multivariate Gaussian distribution. Friedman et al. (2008) used it to analyze a flow cytometry dataset of 11 proteins and 7466 cells, and produced a directed acyclic graph in cell signal data.
In network discovery, Leng and Tang (2012) used it to analyze the U.S. agricultural export data and presented the network and the regions of the US Department of Agriculture export data from 1970 to 2009. Considering that the structure may vary from time to time, Kolar and Xing (2011) proposed that the structure of the undirected graphical model can be consistently estimated in the high-dimensional setting, when the dimensionality of the model is allowed to diverge with the sample size.
7 Experiments
In this section, four numerical experiments are performed to evaluate three of these structured sparse learning methods mentioned in previous sections. The results obtained from these numerical studies are detailed in this section. The first two experiments demonstrate that the structured sparsity-inducing method can recover the signal while the standard sparse learning cannot. The third experiment demonstrates that model parsimony can be obtained through a structured method beyond the standard sparse learning method in wavelet coefficients selection. The fourth experiment demonstrates that the accuracy could be increased with the graphical structure used in logistical regression.
Several publicly accessible software packages are used, including SparseLab (Donoho et al., 2007), SLEP (Liu J et al., 2009), SPAMS (Mairal et al., 2011), YALL1 (Zhang Y et al., 2011), and SPGL1 (van den Berg and Friedlander, 2007). In this article, we use mainly SPAMS and YALL1 to solve the optimization problem. All numerical experiments are conducted with Matlab 7.12.0 on a laptop with an Intel Core I7-4710-MQ, 2.5 GHz CPU, and 4 GB of RAM.
7.1 Measurements
Accuracy, model explainability, and computational complexity are the three most important aspects to consider in machine learning algorithms. The latter two are dominated mainly by model parsimony. Thus, we use prediction accuracy on a test dataset and model parsimony to evaluate the structured sparse learning methods in this study.
7.2 Signal recovery with fused structured sparsity
Fig. 3b shows that the fused model can preserve the natural ordering rather well. Fig. 3c shows that the sparse model presents a sparse solution, but cannot preserve the natural ordering. This example demonstrates that fused structured sparse learning surpasses sparse learning when the original data has a fused structure.
7.3 Signal recovery improved by grouped structured sparsity
To show the capability of the grouped structure for solving grouped structured problems, we conduct the following tests. We consider solving problem (3) with weighted groups. First, we create a random m × n encoding matrix, and perform scaling by normalizing the rows of the encoding matrix. Then, we generate groups with the desired number of unique groups, and a weight is determined for each group. Next, the observations are generated through the encoding matrix and grouped sparse vector, followed by a Gaussian noise added to the observation.
7.4 Model parsimony gained by hierarchically structured sparsity
To show the capability of the hierarchical structure, we conduct the following tests. First, an image with five rectangles is randomly generated. Then, we apply ordinary linear sampling to measure the 4096 wavelet coefficients directly. Compared to the linear structure, we also perform a hierarchical structure that has 1152 wavelet coefficients.
7.5 Prediction accuracy improved by graphically structured sparsity
To show the ability of the graphical structure to solve graphically structured problems, we conduct the following tests. The experiment is conducted on the binary classification datasets: ‘20 Newsgroups’ (www.cs.nyu.edu/~roweis/data.html). The 20 Newsgroups dataset is a collection of approximately 20 000 newsgroup documents, partitioned (nearly) evenly across 20 different newsgroups. The 20 Newsgroups collection has become a popular dataset for experiments in text applications of machine learning techniques, such as text classification and text clustering. Here, the first 100 words are selected and we use 80% of the samples for training and 20% for testing. To reduce statistical variability, experimental results are averaged over 10 repetitions. First, we use graphical Lasso to generate the graphical relationship. Then, we propose a novel graph-guided method to do classification, and compare the average predication accuracy of the standard classifier without a graphical structure penalty.
8 Conclusions and discussion
Structured sparse learning methods incorporate specific structure information with sparse learning methods, and have been used in various fields. In this article, we reviewed the development of the theory, formulations, algorithms, and applications of the latest structured sparse learning methods, including grouped structured sparsity, fused structured sparsity, hierarchically structured sparsity, and graphically structured sparsity. For each type of structured sparsity, we presented the original formulation and its variations and the mathematical motivation of these methods, addressed the algorithms for solving these problems, and discussed the fact that applications with prior knowledge lead to improved explicability of the sparse estimations and/or increased prediction performance in related research fields.
Experiments have been conducted to demonstrate the advantage of structured sparse learning algorithms beyond standard sparse learning methods. These experiments demonstrated that the structured sparsity-inducing method could achieve better performance than the standard sparse learning method. We also proposed a novel graph-guided logistic regression method to demonstrate the efficacy of the graphical structure. However, the experiment results on super computers (Yang et al., 2010; 2011) are expected and power efficient algorithms (Lai et al., 2015; 2016) and algorithms for new infrastructures (Chen et al., 2016) are still required.
- 1.
Online learning algorithms for structured sparsity problems. Most current structured sparsity is optimized in a batch way. In real-world applications, the training data volume may be huge or be given in a sequential way. Online learning is a better choice to address these problems.
- 2.
Efficient algorithms for non-convex models. Today, most structured models are solved through a convex approximation of the original formulation. However, in statistics, it was reported in the literature that non-convex regularization usually yields a solution with more desirable structural properties, for example, the ℓ_{0}-norm regularized least squares problem (i.e., l(·) is a least squares function). Efficient non-convex algorithms are needed for these strict models.
- 3.
Specific structure inducing regularization. Many structure-inducing regularizations have been proposed, and many of them have been applied in a wide range of fields. For specifically structured problems, we still need new regularization to induce the specific structure.
References
- Asaei, A., Bourlard, H., Cevher, V., 2011a. Modelbased compressive sensing for multi-party distant speech recognition. Proc. ICASSP, p.4600–4603. http://dx.doi.org/10.1109/ICASSP.2011.5947379Google Scholar
- Asaei, A., Taghizadeh, M.J., Bourlard, H., et al., 2011b. Multi-party speech recovery exploiting structured sparsity models. Proc. Conf. on Int. Speech Communication Association, p.192–195.Google Scholar
- Asaei, A., Bourlard, H., Taghizadeh, M.J., et al., 2014a. Model-based sparse component analysis for reverberant speech localization. Proc. ICASSP, p.1439–1443 http://dx.doi.org/10.1109/ICASSP.2014.6853835Google Scholar
- Asaei, A., Golbabaee, M., Bourlard, H., et al., 2014b. Structured sparsity models for reverberant speech separation. IEEE/ACM Trans. Audio Speech Lang. Process., 22(3): 620–633. http://dx.doi.org/10.1109/TASLP.2013.2297012Google Scholar
- Bach, F., 2008a. Consistency of trace norm minimization. J. Mach. Learn. Res., 9: 1019–1048.MathSciNetzbMATHGoogle Scholar
- Bach, F., 2008b. Consistency of the group Lasso and multiple kernel learning. J. Mach. Learn. Res., 9: 1179–1225.MathSciNetzbMATHGoogle Scholar
- Bach, F., Jenatton, R., Mairal, J., et al., 2011. Convex optimization with sparsity-inducing norms. In: Sra, S., Nowozin, S., Wright, S.J. (Eds.), Optimization for Machine Learning. MIT Press, Cambridge, p.1–35.zbMATHGoogle Scholar
- Bach, F., Jenatton, R., Mairal, J., et al., 2012a. Optimization with sparsity-inducing penalties. Found. Trends Mach. Learn., 4(1): 1–106. http://dx.doi.org/10.1561/2200000015zbMATHGoogle Scholar
- Bach, F., Jenatton, R., Mairal, J., et al., 2012b. Structured sparsity through convex optimization. Stat. Sci., 27(4): 450–468. http://dx.doi.org/10.1214/12-STS394MathSciNetzbMATHGoogle Scholar
- Bach, F., Jordan, M.I., 2006. Learning spectral clustering, with application to speech separation. J. Mach. Learn. Res., 7: 1963–2001.MathSciNetzbMATHGoogle Scholar
- Banerjee, O., El Ghaoui, L., d’Aspremont, A., 2008. Model selection through sparse maximum likelihood estimation for multivariate Gaussian or binary data. J. Mach. Learn. Res., 9: 485–516.MathSciNetzbMATHGoogle Scholar
- Baraniuk, R.G., Cevher, V., Duarte, M.F., et al., 2010. Model-based compressive sensing. IEEE Trans. Inform. Theory, 56(4): 1982–2001. http://dx.doi.org/10.1109/Tit.2010.2040894MathSciNetzbMATHGoogle Scholar
- Beck, A., Teboulle, M., 2003. Mirror descent and nonlinear projected subgradient methods for convex optimization. Oper. Res. Lett., 31(3): 167–175. http://dx.doi.org/10.1016/S0167-6377(02)00231-6MathSciNetzbMATHGoogle Scholar
- Beck, A., Teboulle, M., 2009. A fast iterative shrinkagethresholding algorithm for linear inverse problems.Google Scholar
- SIAM J. Imag. Sci., 2(1): 183–202. http://dx.doi.org/10.1137/080716542
- Bengio, S., Pereira, F., Singer, Y., et al., 2009. Group sparse coding. Proc. NIPS, p.82–89.Google Scholar
- Blei, D.M., Griffiths, T.L., Jordan, M.I., 2010. The nested Chinese restaurant process and Bayesian nonparametric inference of topic hierarchies. J. ACM, 57(2):7. http://doi.acm.org/10.1145/1667053.1667056MathSciNetzbMATHGoogle Scholar
- Borne, K., 2009. Scientific data mining in astronomy. arXiv:0911.0505.Google Scholar
- Boyd, S., Parikh, N., Chu, E., et al., 2011. Distributed optimization and statistical learning via the alternating direction method of multipliers. Found. Trends Mach. Learn., 3(1): 1–122. http://dx.doi.org/10.1561/2200000016zbMATHGoogle Scholar
- Bruckstein, A.M., Donoho, D.L., Elad, M., 2009. From sparse solutions of systems of equations to sparse modeling of signals and images. SIAM Rev., 51(1): 34–81. http://dx.doi.org/10.1137/060657704MathSciNetzbMATHGoogle Scholar
- Candès, E., Tao, T., 2007. The Dantzig selector: statistical estimation when p is much larger than n. Ann. Stat., 35(6): 2313–2351. http://dx.doi.org/10.1214/009053606000001523MathSciNetzbMATHGoogle Scholar
- Candès, E.J., 2008. The restricted isometry property and its implications for compressed sensing. Comput. Rend. Math., 346(9-10):589–592. http://dx.doi.org/10.1016/j.crma.2008.03.014MathSciNetzbMATHGoogle Scholar
- Candès, E.J., Recht, B., 2009. Exact matrix completion via convex optimization. Found. Comput. Math., 9(6): 717–772. http://dx.doi.org/10.1007/s10208-009-9045-5MathSciNetzbMATHGoogle Scholar
- Candès, E.J., Romberg, J.K., Tao, T., 2006. Stable signal recovery from incomplete and inaccurate measurements. Commun. Pure Appl. Math., 59(8): 1207–1223. http://dx.doi.org/10.1002/Cpa.20124MathSciNetzbMATHGoogle Scholar
- Candès, E.J., Wakin, M.B., Boyd, S.P., 2008. Enhancing sparsity by reweighted 1 minimization. J. Four. Anal. Appl., 14(5): 877–905. http://dx.doi.org/10.1007/s00041-008-9045-xMathSciNetzbMATHGoogle Scholar
- Chandrasekaran, V., Parrilo, P.A., Willsky, A.S., 2012. Latent variable graphical model selection via convex optimization. Ann. Stat., 40(4): 1935–1967. http://dx.doi.org/10.1214/11-AOS949MathSciNetzbMATHGoogle Scholar
- Chartrand, R., Yin, W.T., 2008. Iteratively reweighted algorithms for compressive sensing. Proc. ICASSP, p.3869–3872. http://dx.doi.org/10.1109/Icassp.2008.4518498Google Scholar
- Chen, C., Huang, J.Z., 2014. Exploiting the wavelet structure in compressed sensing MRI. Magn. Reson. Imag., 32(10): 1377–1389. http://dx.doi.org/10.1016/j.mri.2014.07.016Google Scholar
- Chen, C., Li, Y.Q., Huang, J.Z., 2014. Forest sparsity for multi-channel compressive sensing. IEEE Trans. Signal Process., 62(11): 2803–2813. http://dx.doi.org/10.1109/TSP.2014.2318138MathSciNetGoogle Scholar
- Chen, H.Y., Sun, Z.G., Yi, F., et al., 2016. BufferBank storage: an economic, scalable and universally usable innetwork storage model for streaming data applications. Sci. China Inform. Sci., 59(1): 1–15. http://dx.doi.org/10.1007/s11432-015-5299-5Google Scholar
- Chen, S., Donoho, D., 1994. Basis pursuit. Proc. Asilomar Conf. on Signals, Systems and Computers, p.41–44.Google Scholar
- Chen, X., Lin, Q.H., Kim, S., et al., 2012. Smoothing proximal gradient method for general structured sparse regression. Ann. Appl. Stat., 6(2): 719–752. http://dx.doi.org/10.1214/11-AOAS514MathSciNetzbMATHGoogle Scholar
- Combettes, P.L., Pesquet, J.C., 2011. Proximal splitting methods in signal processing. In: Bauschke, H.H., Burachik, R.S., Combettes, P.L., et al. (Eds.), Fixed-Point Algorithms for Inverse Problems in Science and Engineering. Springer, Berlin, p.185–212. http://dx.doi.org/10.1007/978-1-4419-9569-8_10CrossRefzbMATHGoogle Scholar
- Dempster, A.P., 1972. Covariance selection. Biometrics, 28: 157–175.Google Scholar
- Donoho, D.L., Huo, X., 2001. Uncertainty principles and ideal atomic decomposition. IEEE Trans. Inform. Theory, 47(7): 2845–2862. http://dx.doi.org/10.1109/18.959265MathSciNetzbMATHGoogle Scholar
- Donoho, D.L, Drori, I., Stodden, V.C, et al., 2007. Sparse-Lab. http://sparselab.stanford.edu/
- Duarte, M.F., Eldar, Y.C., 2011. Structured compressed sensing: from theory to applications. IEEE Trans. Signal Process., 59(9): 4053–4085. http://dx.doi.org/10.1109/TSP.2011.2161982MathSciNetGoogle Scholar
- Elad, M., 2010. Sparse and Redundant Representations: from Theory to Applications in Signal and Image Processing. Springer, Berlin. http://dx.doi.org/10.1007/978-1-4419-7011-4CrossRefzbMATHGoogle Scholar
- Fan, J.Q., Li, R.Z., 2011. Variable selection via nonconcave penalized likelihood and its oracle properties. J. Am. Stat. Assoc., 96(456): 1348–1360. http://dx.doi.org/10.1198/016214501753382273MathSciNetzbMATHGoogle Scholar
- Fan, J.Q., Lv, J.C., Qi, L., 2011. Sparse high-dimensional models in economics. Ann. Rev. Econ., 3: 291–317. http://dx.doi.org/10.1146/annurev-economics-061109-080451Google Scholar
- Foucart, S., Lai, M.J., 2009. Sparsest solutions of underdetermined linear systems via lq-minimization for 0 < q = 1. Appl. Comput. Harmon. Anal., 26(3): 395–407. http://dx.doi.org/10.1016/j.acha.2008.09.001MathSciNetzbMATHGoogle Scholar
- Friedman, J., Hastie, T., Höfling, H., et al., 2007. Pathwise coordinate optimization. Ann. Appl. Stat., 1(2): 302–332. http://dx.doi.org/10.1214/07-Aoas131MathSciNetzbMATHGoogle Scholar
- Friedman, J., Hastie, T., Tibshirani, R., 2008. Sparse inverse covariance estimation with the graphical Lasso. Biostatistics, 9(3): 432–441. http://dx.doi.org/10.1093/biostatistics/kxm045zbMATHGoogle Scholar
- Garber, D., Hazan, E., 2015. Faster rates for the Frank-Wolfe method over strongly-convex sets. Proc. ICML, p.541–549.Google Scholar
- Gill, P.E., Murray, W., Saunders, M.A., 2008. User’s Guide for SQOPT Version 7: Software for Large-Scale Linear and Quadratic Programming. http://www-leland.stanford.edu/group/SOL/guides/sqdoc7.pdfGoogle Scholar
- Gong, P.H., Zhang, C.S., Lu, Z.S., et al., 2013. A general iterative shrinkage and thresholding algorithm for non-convex regularized optimization problems. Proc. ICML, p.37–45.Google Scholar
- Grant, M., Boyd, S., 2013. CVX: Matlab Software for Disciplined Convex Programming. Version 2.0 Beta. http://cvxr.com/cvx/
- Hazan, E., Agarwal, A., Kale, S., 2007. Logarithmic regret algorithms for online convex optimization. Mach. Learn., 69(2): 169–192. http://dx.doi.org/10.1007/s10994-007-5016-8zbMATHGoogle Scholar
- Hoefling, H., 2010. A path algorithm for the fused Lasso signal approximator. J. Comput. Graph. Stat., 19(4): 984–1006. http://dx.doi.org/10.1198/jcgs.2010.09208MathSciNetGoogle Scholar
- Hong, M.Y., Razaviyayn, M., Luo, Z.Q., et al., 2015. A unified algorithmic framework for block-structured optimization involving big data. arXiv:1511.02746.Google Scholar
- Hu, T.C., Yu, J.H., 2016. Max-margin based Bayesian classifier. Front. Inform. Technol. Electron. Eng., 17(10): 973–981. http://dx.doi.org/10.1631/FITEE.1601078Google Scholar
- Huang, J.Z., Zhang, T., Metaxas, D., 2011. Learning with structured sparsity. J. Mach. Learn. Res., 12: 3371–3412.MathSciNetzbMATHGoogle Scholar
- Huang, T., Wu, B.L., Lizardi, P., et al., 2005. Detection of DNA copy number alterations using penalized least squares regression. Bioinformatics, 21(20): 3811–3817. https://doi.org/10.1093/bioinformatics/bti646Google Scholar
- Jacob, L., Obozinski, G., Vert, J.P., 2009. Group Lasso with overlap and graph Lasso. Proc. ICML, p.433–440. http://dx.doi.org/10.1145/1553374.1553431
- Jaggi, M., 2013. Revisiting Frank-Wolfe: projection-free sparse convex optimization. Proc. ICML, p.427–435.Google Scholar
- Jenatton, R., 2011. Structured Sparsity-Inducing Norms: Statistical and Algorithmic Properties with Applications to Neuroimaging. PhD Thesis, École Normale Supérieure de Cachan, Cachan, France.Google Scholar
- Jenatton, R., Obozinski, G., Bach, F., 2009. Structured sparse principal component analysis. Proc. AISTATS, p.366–373.Google Scholar
- Jenatton, R., Mairal, J., Bach, F.R., et al., 2010. Proximal methods for sparse hierarchical dictionary learning. Proc. ICML, p.487–494.Google Scholar
- Jenatton, R., Mairal, J., Obozinski, G., et al., 2011. Proximal methods for hierarchical sparse coding. J. Mach. Learn. Res., 12: 2297–2334.MathSciNetzbMATHGoogle Scholar
- Jenatton, R., Gramfort, A., Michel, V., et al., 2012. Multiscale mining of fMRI data with hierarchical structured sparsity.Google Scholar
- SIAM J. Imag. Sci., 5(3): 835–856. http://dx.doi.org/10.1137/110832380
- John Lu, Z.Q., 2010. The elements of statistical learning: data mining, inference, and prediction. J. R. Stat. Soc. A, 173(3): 693–694. http://dx.doi.org/10.1111/j.1467-985X.2010.00646_6.xGoogle Scholar
- Jones, B., West, M., 2005. Covariance decomposition in undirected Gaussian graphical models. Biometrika, 92(4): 779–786. https://doi.org/10.1093/biomet/92.4.779MathSciNetzbMATHGoogle Scholar
- Karygianni, S., Frossard, P., 2014. Structured sparse coding for image denoising or pattern detection. Proc. ICASSP, p.3533–3537. http://dx.doi.org/10.1109/ICASSP.2014.6854258Google Scholar
- Kim, B.S., Park, J.Y., Gilbert, A.C., et al., 2013. Hierarchical classification of images by sparse approximation. Image Vis. Comput., 31(12): 982–991. http://dx.doi.org/10.1016/j.imavis.2013.10.005Google Scholar
- Kim, S., Xing, E.P., 2010. Tree-guided group Lasso for multi-task regression with structured sparsity. Proc. ICML, p.543–550.Google Scholar
- Kim, S., Xing, E.P., 2012. Tree-guided group Lasso for multi-response regression with structured sparsity, with an application to eQTL mapping. Ann. Appl. Stat., 6(3): 1095–1117. http://dx.doi.org/10.1214/12-Aoas549MathSciNetzbMATHGoogle Scholar
- Kim, S., Xing, E.P., 2014. Exploiting genome structure in association analysis. J. Comput. Biol., 21(4): 345–360. http://dx.doi.org/10.1089/cmb.2009.0224MathSciNetGoogle Scholar
- Kolar, M., Xing, E.P., 2011. On time varying undirected graphs. Proc. AISTATS, p.407–415.Google Scholar
- Koren, Y., Bell, R., Volinsky, C., 2009. Matrix factorization techniques for recommender systems. Computer, 42(8): 30–37. http://dx.doi.org/10.1109/MC.2009.263Google Scholar
- Lacoste-Julien, S., Schmidt, M., Bach, F., 2012. A simpler approach to obtaining an O(1/t) convergence rate for the projected stochastic subgradient method. arXiv:1212.2002.Google Scholar
- Lai, M.J., Xu, Y.Y., Yin, W.T., 2013. Improved iteratively reweighted least squares for unconstrained smoothed lq minimization. SIAM J. Numer. Anal., 51(2): 927–957. http://dx.doi.org/10.1137/110840364MathSciNetzbMATHGoogle Scholar
- Lai, Z.Q., Lam, K.T., Wang, C.L., et al., 2015. Latencyaware DVFS for efficient power state transitions on many-core architectures. J. Supercomput., 71(7): 2720–2747. http://dx.doi.org/10.1007/s11227-015-1415-yGoogle Scholar
- Lai, Z.Q., Lam, K.T., Wang, C.L., et al., 2016. PoweRock: power modeling and flexible dynamic power management for many-core architectures. IEEE Syst. J., in press. http://dx.doi.org/10.1109/JSYST.2015.2499307
- Leng, C.L., Tang, C.Y., 2012. Sparse matrix graphical models. J. Am. Stat. Assoc., 107(499): 1187–1200. http://dx.doi.org/10.1080/01621459.2012.706133MathSciNetzbMATHGoogle Scholar
- Li, X.X., Mo, L.L., Yuan, X.M., et al., 2014. Linearized alternating direction method of multipliers for sparse group and fused Lasso models. Comput. Stat. Data Anal., 79: 203–221. http://dx.doi.org/10.1016/j.csda.2014.05.017MathSciNetGoogle Scholar
- Lin, H.Z., Mairal, J.L., Harchaoui, Z., 2015. A universal catalyst for first-order optimization. Proc. NIPS, p.3384–3392.Google Scholar
- Liu, H., Palatucci, M., Zhang, J., 2009. Blockwise coordinate descent procedures for the multi-task Lasso, with applications to neural semantic basis discovery. Proc. ICML, p.649–656. http://dx.doi.org/10.1145/1553374.1553458Google Scholar
- Liu, J., Ji, S., Ye, J., 2009. SLEP: Sparse Learning with Efficient Projections. http://www.public.asu.edu/~jye02/Software/SLEP
- Ma, S.Q., Xue, L.Z., Zou, H., 2013. Alternating direction methods for latent variable Gaussian graphical model selection. Neur. Comput., 25(8): 2172–2198. http://dx.doi.org/10.1162/NECO_a_00379MathSciNetGoogle Scholar
- Mairal, J., 2013. Optimization with first-order surrogate functions. Proc. ICML, p.783–791.Google Scholar
- Mairal, J., Bach, F., Ponce, J., et al., 2011. SPAMS: SPArse Modeling Software. http://spams-devel.gforge.inria.fr/
- Mairal, J., Bach, F., Ponce, J., 2014. Sparse modeling for image and vision processing. Found. Trends Comput. Graph. Vis., 8(2-3):85–283. http://dx.doi.org/10.1561/0600000058zbMATHGoogle Scholar
- Mallat, S., 2008. A Wavelet Tour of Signal Processing: the Sparse Way (3rd Ed.). Elsevier/Academic Press, Amsterdam.zbMATHGoogle Scholar
- McAuley, J., Ming, J., Stewart, D., et al., 2005. Subband correlation and robust speech recognition. IEEE Trans. Speech Audio Process., 13(5): 956–964. http://dx.doi.org/10.1109/TSA.2005.851952Google Scholar
- Meier, L., van de Geer, S., Bühlmann, P., 2008. The group Lasso for logistic regression. J. R. Stat. Soc. B, 70(1): 53–71. http://dx.doi.org/10.1111/j.1467-9868.2007.00627.xMathSciNetzbMATHGoogle Scholar
- Meinshausen, N., Bühlmann, P., 2006. High-dimensional graphs and variable selection with the Lasso. Ann. Stat., 34(3): 1436–1462. http://dx.doi.org/10.1214/009053606000000281MathSciNetzbMATHGoogle Scholar
- Meinshausen, N., Yu, B., 2008. Lasso-type recovery of sparse representations for high-dimensional data. Ann. Stat., 37(1): 246–270. http://dx.doi.org/10.1214/07-AOS582MathSciNetzbMATHGoogle Scholar
- Micchelli, C.A., Morales, J.M., Pontil, M., 2013. Regularizers for structured sparsity. Adv. Comput. Math., 38(3): 455–489. http://dx.doi.org/10.1007/s10444-011-9245-9MathSciNetzbMATHGoogle Scholar
- Mosci, S., Rosasco, L., Santoro, M., et al., 2010. Solving structured sparsity regularization with proximal methods. LNCS, 6322: 418–433. http://dx.doi.org/10.1007/978-3-642-15883-4_27Google Scholar
- Mougeot, M., Picard, D., Tribouley, K., 2013. Grouping strategies and thresholding for high dimensional linear models. J. Stat. Plan. Infer., 143(9): 1417–1438. http://dx.doi.org/10.1016/j.jspi.2013.03.001MathSciNetzbMATHGoogle Scholar
- Najafian, M., 2016. Acoustic Model Selection for Recognition of Regional Accented Speech. PhD Thesis, University of Birmingham, Birmingham, UK.Google Scholar
- Negahban, S.N., Ravikumar, P., Wainwright, M.J., et al., 2012. A unified framework for high-dimensional analysis of m-estimators with decomposable regularizers. Stat. Sci., 27(4): 538–557. http://dx.doi.org/10.1214/12-Sts400MathSciNetzbMATHGoogle Scholar
- Nemirovski, A., 2004. Prox-method with rate of convergence O(1/t) for variational inequalities with Lipschitz continuous monotone operators and smooth convex-concave saddle point problems.Google Scholar
- SIAM J. Optim., 15(1): 229–251. http://dx.doi.org/10.1137/S1052623403425629
- Nesterov, Y., 2004. Introductory Lectures on Convex Optimization: a Basic Course. Springer Science & Business Media. http://dx.doi.org/10.1007/978-1-4419-8853-9CrossRefzbMATHGoogle Scholar
- Nesterov, Y., 2009. Primal-dual subgradient methods for convex problems. Math. Program., 120(1): 221–259. http://dx.doi.org/10.1007/s10107-007-0149-xMathSciNetzbMATHGoogle Scholar
- Parikh, N., Boyd, S., 2014. Proximal algorithms. Found. Trends Optim., 1(3): 127–239. http://dx.doi.org/10.1561/2400000003Google Scholar
- Peng, Z.M., Wu, T.Y., Xu, Y.Y., et al., 2016. Coordinate friendly structures, algorithms and applications. arXiv:1601.00863.zbMATHGoogle Scholar
- Qiao, L.B., Lin, T.Y., Jiang, Y.G., et al., 2016a. On stochastic primal-dual hybrid gradient approach for compositely regularized minimization. Proc. European Conf. on Artificial Intelligence, p.167–174. http://dx.doi.org/10.3233/978-1-61499-672-9-167
- Qiao, L.B., Zhang, B.F., Su, J.S., et al., 2016b. Linearized alternating direction method of multipliers for constrained nonconvex regularized optimization. Proc. Asian Conf. on Machine Learning, p.97–109.Google Scholar
- Qiao, L.B., Zhang, B.F., Zhuang, L., et al., 2016c. An efficient algorithm for tensor principal component analysis via proximal linearized alternating direction method of multipliers. Proc. Int. Conf. on Advanced Cloud and Big Data, p.283–288. http://dx.doi.org/10.1109/CBD.2016.056Google Scholar
- Rakotomamonjy, A., 2011. Surveying and comparing simultaneous sparse approximation (or group-Lasso) algorithms. Signal Process., 91(7): 1505–1526. http://dx.doi.org/10.1016/j.sigpro.2011.01.012zbMATHGoogle Scholar
- Rasmussen, C.E., Ghahramani, Z., 2001. Occam’s razor. Proc. NIPS, p.294–300.Google Scholar
- Rendle, S., Schmidt-Thieme, L., 2010. Pairwise interaction tensor factorization for personalized tag recommendation. Proc. 3rd ACM Int. Conf. on Web Wearch and Data Mining, p.81–90. http://dx.doi.org/10.1145/1718487.1718498Google Scholar
- Roth, V., Fischer, B., 2008. The group-Lasso for generalized linear models: uniqueness of solutions and efficient algorithms. Proc. ICML, p.848–855. http://dx.doi.org/10.1145/1390156.1390263CrossRefGoogle Scholar
- Rudin, L.I., Osher, S., Fatemi, E., 1992. Nonlinear total variation based noise removal algorithms. Phys. D, 60(1-4):259–268. http://dx.doi.org/10.1016/0167-2789(92)90242-FMathSciNetzbMATHGoogle Scholar
- Scheinberg, K., Ma, S., Goldfarb, D., 2010. Sparse inverse covariance selection via alternating linearization methods. Proc. NIPS, p.2101–2109.Google Scholar
- Selesnick, I.W., Bayram, I., 2014. Sparse signal estimation by maximally sparse convex optimization. IEEE Trans. Signal Process., 62(5): 1078–1092. http://dx.doi.org/10.1109/TSP.2014.2298839MathSciNetGoogle Scholar
- Simon, N., Friedman, J., Hastie, T., et al., 2013. A sparsegroup Lasso. J. Comput. Graph. Stat., 22(2): 231–245. http://dx.doi.org/10.1080/10618600.2012.681250Google Scholar
- Su, W.J., Boyd, S., Candès, E., 2014. A differential equation for modeling Nesterov’s accelerated gradient method: theory and insights. Proc. NIPS, p.2510–2518.Google Scholar
- Sun, Y.P., Chen, S.H., Han, B., et al., 2015a. A novel location privacy mining threat in vehicular Internet access service. LNCS, 9204: 467–476. http://dx.doi.org/10.1007/978-3-319-21837-3_46Google Scholar
- Sun, Y.P., Zhang, B.F., Zhao, B.K., et al., 2015b. Mix-zones optimal deployment for protecting location privacy in VANET. Peer-to-Peer Netw. Appl., 8(6): 1108–1121. http://dx.doi.org/10.1007/s12083-014-0269-zGoogle Scholar
- Suzuki, T.J., 2013. Dual averaging and proximal gradient descent for online alternating direction multiplier method. Proc. ICML, p.392–400.Google Scholar
- Takacs, G., Pilaszy, I., Nemeth, B., et al., 2009. Scalable collaborative filtering approaches for large recommender systems. J. Mach. Learn. Res., 10: 623–656.Google Scholar
- Tibshirani, R., 1996. Regression shrinkage and selection via the Lasso. J. R. Stat. Soc. B, 58(1): 267–288.MathSciNetzbMATHGoogle Scholar
- Tibshirani, R., Wang, P., 2008. Spatial smoothing and hot spot detection for CGH data using the fused Lasso. Biostatistics, 9(1): 18–29. http://dx.doi.org/10.1093/biostatistics/kxm013zbMATHGoogle Scholar
- Tibshirani, R., Saunders, M., Rosset, S., et al., 2005. Sparsity and smoothness via the fused Lasso. J. R. Stat. Soc. B, 67(1): 91–108. http://dx.doi.org/10.1111/j.1467-9868.2005.00490.xMathSciNetzbMATHGoogle Scholar
- Toh, K., Todd, M.J., Tütüncü, R.H., 2006. SDPT3 Version 4.0: a Matlab Software for Semidefinite-Quadratic-Linear Programming. http://www.math.nus.edu.sg/~mattohkc/sdpt3.html
- Tropp, J.A., 2004. Greed is good: algorithmic results for sparse approximation. IEEE Trans. Inform. Theory, 50(10): 2231–2242. http://dx.doi.org/10.1109/Tit.2004.834793MathSciNetzbMATHGoogle Scholar
- Tropp, J.A., Gilbert, A.C., Muthukrishnan, S., et al., 2003. Improved sparse approximation over quasi-incoherent dictionaries. Proc. Int. Conf. on Image Processing, p.37–40. http://dx.doi.org/10.1109/ICIP.2003.1246892Google Scholar
- Tseng, P., 2008. On Accelerated Proximal Gradient Methods for Convex-Concave Optimization. http://www.mit.edu/~dimitrib/PTseng/papers/apgm.pdf
- Tseng, P., Yun, S., 2009. A coordinate gradient descent method for nonsmooth separable minimization. Math. Program., 117(1): 387–423. http://dx.doi.org/10.1007/s10107-007-0170-0MathSciNetzbMATHGoogle Scholar
- van den Berg, E., Friedlander, M.P., 2007. SPGL1: a Solver for Large-Scale Sparse Reconstruction. http://www.cs.ubc.ca/labs/scl/spgl1
- Villa, S., Rosasco, L., Mosci, S., et al., 2014. Proximal methods for the latent group Lasso penalty. Compt. Optim. Appl., 58(2): 381–407. http://dx.doi.org/10.1007/s10589-013-9628-6MathSciNetzbMATHGoogle Scholar
- Vincent, M., Hansen, N.R., 2014. Sparse group Lasso and high dimensional multinomial classification. Comput. Stat. Data Anal., 71: 771–786. http://dx.doi.org/10.1016/j.csda.2013.06.004MathSciNetGoogle Scholar
- Wainwright, M.J., Jordan, M.I., 2008. Graphical models, exponential families, and variational inference. Found. Trend. Mach. Learn., 1(1-2):1–305. http://dx.doi.org/10.1561/2200000001zbMATHGoogle Scholar
- Wang, H.S., Leng, C.L., 2008. A note on adaptive group Lasso. Comput. Stat. Data Anal., 52(12): 5277–5286. http://dx.doi.org/10.1016/j.csda.2008.05.006MathSciNetzbMATHGoogle Scholar
- Wang, L.C., You, Y., Lian, H., 2013. A simple and efficient algorithm for fused Lasso signal approximator with convex loss function. Comput. Stat., 28(4): 1699–1714. http://dx.doi.org/10.1007/s00180-012-0373-6MathSciNetzbMATHGoogle Scholar
- Wang, Y., Wang, J.J., Xu, Z.B., 2013. On recovery of blocksparse signals via mixed 2/q (0 < q = 1) norm minimization. EURASIP J. Adv. Signal Process., 2013:1–17. http://dx.doi.org/10.1186/1687-6180-2013-76Google Scholar
- Wen, Z., Goldfarb, D., Scheinberg, K., 2012. Block coordinate descent methods for semidefinite programming. In: Anjos, M.F., Lasserre, J.B. (Eds.), Handbook on Semidefinite, Conic and Polynomial Optimization. Springer US, Boston, p.533–564. http://dx.doi.org/10.1007/978-1-4614-0769-0_19CrossRefzbMATHGoogle Scholar
- Wermuth, N., 1976. Analogies between multiplicative models for contingency tables and covariance selection. Biometrics, 32: 95–108.MathSciNetzbMATHGoogle Scholar
- Wille, A., Bü hlmann, P., 2006. Low-order conditional independence graphs for inferring genetic networks. Stat. Appl. Genet. Mol. Biol., 5(1). http://dx.doi.org/10.2202/1544-6115.1170
- Wrinch, D., Jeffreys, H., 1921. On certain fundamental principles of scientific inquiry. Phil. Mag., 42(249): 369–390. http://dx.doi.org/10.1080/14786442108633773zbMATHGoogle Scholar
- Wu, Y.L., Lu, X.C., Su, J.S., et al., 2016. An efficient searchable encryption against keyword guessing attacks for sharable electronic medical records in cloud-based system. J. Med. Syst., 40:258. http://dx.doi.org/10.1007/s10916-016-0609-zGoogle Scholar
- Xiao, J.J., Qiao, L.B., Stolkin, R., et al., 2016. Distractorsupported single target tracking in extremely cluttered scenes. LNCS, 9908: 121–136. http://dx.doi.org/10.1007/978-3-319-46493-0_8Google Scholar
- Xiao, L., Zhang, T., 2014. A proximal stochastic gradient method with progressive variance reduction.Google Scholar
- SIAM J. Optim., 24(4): 2057–2075. http://dx.doi.org/10.1137/140961791
- Xie, H., Tong, R.F., 2016. Image meshing via hierarchical optimization. Front. Inform. Technol. Electron. Eng., 17(1): 32–40. http://dx.doi.org/10.1631/FITEE.1500171Google Scholar
- Xie, Y.C., Huang, H., Hu, Y., et al., 2016. Applications of advanced control methods in spacecrafts: progress, challenges, and future prospects. Front. Inform. Technol. Electron. Eng., 17(9): 841–861. http://dx.doi.org/10.1631/FITEE.1601063Google Scholar
- Xie, Z.X., Xu, Y., 2014. Sparse group Lasso based uncertain feature selection. Int. J. Mach. Learn. Cybern., 5(2): 201–210. http://dx.doi.org/10.1007/s13042-013-0156-6Google Scholar
- Xu, X., Zhang, B.F., Zhong, Q.X., 2005. Text categorization using SVMs with Rocchio ensemble for Internet information classification. LNCS, 3619: 1022–1031. http://dx.doi.org/10.1007/11534310_107Google Scholar
- Xu, X., Hu, D.W., Lu, X.C., 2007. Kernel-based least squares policy iteration for reinforcement learning. IEEE Trans. Neur. Netw., 18(4): 973–992. http://dx.doi.org/10.1109/tnn.2007.899161Google Scholar
- Xu, X., Liu, C.M., Yang, S.X., et al., 2011. Hierarchical approximate policy iteration with binary-tree state space decomposition. IEEE Trans. Neur. Netw., 22(12): 1863–1877. http://dx.doi.org/10.1109/tnn.2011.2168422Google Scholar
- Xu, Z., Chang, X., Xu, F., et al., 2012. L1/2 regularization: a thresholding representation theory and a fast solver. IEEE Trans. Neur. Netw. Learn. Syst., 23(7): 1013–1027. http://dx.doi.org/10.1109/TNNLS.2012.2197412Google Scholar
- Yang, J.F., Yuan, X.M., 2013. Linearized augmented Lagrangian and alternating direction methods for nuclear norm minimization. Math. Comput., 82: 301–329. http://dx.doi.org/10.1090/S0025-5718-2012-02598-1MathSciNetzbMATHGoogle Scholar
- Yang, X.J., Liao, X.K., Xu, W.X., et al., 2010. Th-1: China’s first petaflop supercomputer. Front. Comput. Sci. China, 4(4): 445–455. http://dx.doi.org/10.1007/s11704-010-0383-xGoogle Scholar
- Yang, X.J., Liao, X.K., Lu, K., et al., 2011. The TianHe-1A supercomputer: its hardware and software. J. Comput. Sci. Technol., 26(3): 344–351. http://dx.doi.org/10.1007/s11390-011-1137-4Google Scholar
- Ye, G.B., Xie, X.H., 2011. Split Bregman method for large scale fused Lasso. Comput. Stat. Data Anal., 55(4): 1552–1569. http://dx.doi.org/10.1016/j.csda.2010.10.021MathSciNetzbMATHGoogle Scholar
- Yuan, M., Lin, Y., 2006. Model selection and estimation in regression with grouped variables. J. R. Stat. Soc. B, 68(1): 49–67. http://dx.doi.org/10.1111/j.1467-9868.2005.00532.xMathSciNetzbMATHGoogle Scholar
- Yuan, M., Lin, Y., 2007. Model selection and estimation in the Gaussian graphical model. Biometrika, 94(1): 19–35. http://dx.doi.org/10.1093/biomet/asm018MathSciNetzbMATHGoogle Scholar
- Yuan, M., Yang, B.X., Ma, Y.D., et al., 2015. Multi-scale UDCT dictionary learning based highly undersampled MRimage reconstruction using patch-based constraint splitting augmented Lagrangian shrinkage algorithm. Front. Inform. Technol. Electron. Eng., 16(12): 1069–1087. http://dx.doi.org/10.1631/FITEE.1400423Google Scholar
- Zhang, B.F., Su, J.S., Xu, X., 2006. A class-incremental learning method for multi-class support vector machines in text classification. Proc. ICMLC, p.2581–2585. http://dx.doi.org/10.1109/ICMLC.2006.258853Google Scholar
- Zhang, C.H., 2010. Nearly unbiased variable selection under minimax concave penalty. Ann. Stat., 38(2): 894–942. http://dx.doi.org/10.1214/09-AOS729MathSciNetzbMATHGoogle Scholar
- Zhang, S.Z., Wang, K., Chen, B.L., et al., 2011. A new framework for co-clustering of gene expression data. LNCS, 7036: 1–12. http://dx.doi.org/10.1007/978-3-642-24855-9_1Google Scholar
- Zhang, T., 2009. Some sharp performance bounds for least squares regression with L1 regularization. Ann. Stat., 37(5A):2109–2144. http://dx.doi.org/10.1214/08-AOS659zbMATHGoogle Scholar
- Zhang, T., 2010. Analysis of multi-stage convex relaxation for sparse regularization. J. Mach. Learn. Res., 11: 1081–1107.MathSciNetzbMATHGoogle Scholar
- Zhang, T., 2013. Multi-stage convex relaxation for feature selection. Bernoulli, 19(5B):2277–2293. http://dx.doi.org/10.3150/12-BEJ452MathSciNetzbMATHGoogle Scholar
- Zhang, T.Z., Ghanem, B., Liu, S., et al., 2012. Robust visual tracking via multi-task sparse learning. Proc. CVPR, p.2042–2049. http://dx.doi.org/10.1109/CVPR.2012.6247908Google Scholar
- Zhang, T.Z., Ghanem, B., Liu, S., et al., 2013. Robust visual tracking via structured multi-task sparse learning. Int. J. Comput. Vis., 101(2): 367–383. http://dx.doi.org/10.1007/s11263-012-0582-zMathSciNetGoogle Scholar
- Zhang, T.Z., Jia, K., Xu, C.S., et al., 2014. Partial occlusion handling for visual tracking via robust part matching. Proc. CVPR, p.1258–1265. http://dx.doi.org/10.1109/CVPR.2014.164Google Scholar
- Zhang, T.Z., Liu, S., Ahuja, N., et al., 2015a. Robust visual tracking via consistent low-rank sparse learning. Int. J. Comput. Vis., 111(2): 171–190. http://dx.doi.org/10.1007/s11263-014-0738-0Google Scholar
- Zhang, T.Z., Liu, S., Xu, C.S., et al., 2015b. Structural sparse tracking. Proc. CVPR, p.150–158. http://dx.doi.org/10.1109/CVPR.2015.7298610Google Scholar
- Zhang, Y., Yang, J., Yin, W., 2011. YALL1: Your Algorithms for L1. http://yall1.blogs.rice.edu
- Zhang, Z.K., Zhou, T., Zhang, Y.C., 2011. Tag-aware recommender systems: a state-of-the-art survey. J. Comput. Sci. Technol., 26: 767–777. http://dx.doi.org/10.1007/s11390-011-0176-1MathSciNetGoogle Scholar
- Zhao, P., Yu, B., 2006. On model selection consistency of Lasso. J. Mach. Learn. Res., 7: 2541–2563.MathSciNetzbMATHGoogle Scholar
- Zhao, P., Yu, B., 2007. Stagewise Lasso. J. Mach. Learn. Res., 8: 2701–2726.MathSciNetzbMATHGoogle Scholar
- Zhao, P., Rocha, G., Yu, B., 2009. The composite absolute penalties family for grouped and hierarchical variable selection. Ann. Stat., 37(6a):3468–3497. http://dx.doi.org/10.1214/07-Aos584MathSciNetzbMATHGoogle Scholar
- Zhu, Y.T., Zhao, Y.B., Liu, J., et al., 2016. Low complexity robust adaptive beamforming for general-rank signal model with positive semidefinite constraint. Front. Inform. Technol. Electron. Eng., 17(11): 1245–1252. http://dx.doi.org/10.1631/FITEE.1601112Google Scholar