The heavy-tailed distributions of corrupted outliers and singular values of all channels in low-level vision have proven effective priors for many applications such as background modeling, photometric stereo and image alignment. And they can be well modeled by a hyper-Laplacian. However, the use of such distributions generally leads to challenging non-convex, non-smooth and non-Lipschitz problems, and makes existing algorithms very slow for large-scale applications. Together with the analytic solutions to `p-norm minimization with two specific values of p, i.e., p = 1/2 and p = 2/3, we propose two novel bilinear factor matrix norm minimization models for robust principal component analysis. We first define the double nuclear norm and Frobenius/nuclear hybrid norm penalties, and then prove that they are in essence the Schatten-1/2 and 2/3 quasi-norms, respectively, which lead to much more tractable and scalable Lipschitz optimization problems. Our experimental analysis shows that both our methods yield more accurate solutions than original Schatten quasi-norm minimization, even when the number of observations is very limited. Finally, we apply our penalties to various low-level vision problems, e.g., text removal, moving object detection, image alignment and inpainting, and show that our methods usually outperform the state-of-the-art methods.
Recently, nuclear norm-based low rank representation (LRR) methods have been popular in several applications, such as subspace segmentation. However, there exist two limitations: one is that nuclear norm as the relaxation of rank function will lead to the suboptimal solution since nuclear norm-based minimization subproblem tends to the over-relaxations of singular value elements and treats each of them equally; the other is that solving LRR problems may cause more time consumption due to involving singular value decomposition of the large scale matrix at each iteration. To overcome both disadvantages, this paper mainly considers two tractable variants of LRR: one is Schatten-p norm minimization-based LRR (i.e., SpNM_LRR) and the other is Schatten-p norm factorization-based LRR (i.e., SpNFLRR) for p=1, 2/3 and 1/2. By introducing two or more auxiliary variables in the constraints, the alternating direction method of multiplier (ADMM) with multiple updating variables can be devised to solve these variants of LRR. Furthermore, both computational complexity and convergence property are given to evaluate nonconvex multiblocks ADMM algorithms. Several experiments finally validate the efficacy and efficiency of our methods on both synthetic data and real world data.
30 October 2018 in IEEE Transactions on Knowledge and Data Engineering
In this paper, we propose a simple variant of the original SVRG, called variance reduced stochastic gradient descent (VR-SGD). Unlike the choices of snapshot and starting points in SVRG and Prox-SVRG, the two vectors of VR-SGD are set to the average and last iterate of the previous epoch, respectively. The settings allow us to use much larger learning rates, and also make our convergence analysis more challenging. We design two different update rules for smooth and non-smooth problems, respectively, which means that VR-SGD can tackle non-smooth and/or non-strongly convex problems directly without any reduction techniques. Moreover, we analyze the convergence properties of VR-SGD for strongly convex problems, which show that VR-SGD attains linear convergence. Different from most algorithms that have no convergence guarantees for non-strongly convex problems, we also provide the convergence guarantees of VR-SGD for this case, and empirically verify that VR-SGD achieves similar performance to its momentum accelerated variant that has the optimal convergence rate $O(1/T^2)$ . Finally, we apply VR-SGD to solve various machine learning problems, such as empirical risk minimization and leading eigenvalue computation. Experimental results show that VR-SGD converges significantly faster than SVRG and Prox-SVRG, and usually outperforms state-of-the-art accelerated methods, e.g., Katyusha.
3 October 2019 in IEEE Transactions on Neural Networks and Learning Systems
Semi-supervised non-negative matrix factorization (NMF) exploits the strengths of NMF in effectively learning local information contained in data and is also able to achieve effective learning when only a small fraction of data is labeled. NMF is particularly useful for dimensionality reduction of high-dimensional data. However, the mapping between the low-dimensional representation, learned by semi-supervised NMF, and the original high-dimensional data contains complex hierarchical and structural information, which is hard to extract by using only single-layer clustering methods. Therefore, in this article, we propose a new deep learning method, called semi-supervised graph regularized deep NMF with bi-orthogonal constraints (SGDNMF). SGDNMF learns a representation from the hidden layers of a deep network for clustering, which contains varied and unknown attributes. Bi-orthogonal constraints on two factor matrices are introduced into our SGDNMF model, which can make the solution unique and improve clustering performance. This improves the effect of dimensionality reduction because it only requires a small fraction of data to be labeled. In addition, SGDNMF incorporates dual-hypergraph Laplacian regularization, which can reinforce high-order relationships in both data and feature spaces and fully retain the intrinsic geometric structure of the original data. This article presents the details of the SGDNMF algorithm, including the objective function and the iterative updating rules. Empirical experiments on four different data sets demonstrate state-of-the-art performance of SGDNMF in comparison with six other prominent algorithms.
8 June 2020 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Recently, many stochastic variance reduced alternating direction methods of multipliers (ADMMs) (e.g., SAG-ADMM and SVRG-ADMM) have made exciting progress such as linear convergence rate for strongly convex (SC) problems. However, their best-known convergence rate for non-strongly convex (non-SC) problems is $\mathcal {O}(1/T)$ as opposed to $\mathcal {O}(1/T^2)$ of accelerated deterministic algorithms, where $T$ is the number of iterations. Thus, there remains a gap in the convergence rates of existing stochastic ADMM and deterministic algorithms. To bridge this gap, we introduce a new momentum acceleration trick into stochastic variance reduced ADMM, and propose a novel accelerated SVRG-ADMM method (called ASVRG-ADMM) for the machine learning problems with the constraint $Ax + By = c$ . Then we design a linearized proximal update rule and a simple proximal one for the two classes of ADMM-style problems with $B = \tau I$ and $B\ne \tau I$ , respectively, where $I$ is an identity matrix and $\tau$ is an arbitrary bounded constant. Note that our linearized proximal update rule can avoid solving sub-problems iteratively. Moreover, we prove that ASVRG-ADMM converges linearly for SC problems. In particular, ASVRG-ADMM improves the convergence rate from $\mathcal {O}(1/T)$ to $\mathcal {O}(1/T^2)$ for non-SC problems. Finally, we apply ASVRG-ADMM to various machine learning problems, e.g., graph-guided fused Lasso, graph-guided logistic regression, graph-guided SVM, generalized graph-guided fused Lasso and multi-task learning, and show that ASVRG-ADMM consistently converges faster than the state-of-the-art methods.