In this paper, we study a novel transfer learning problem termed Distant Domain Transfer Learning (DDTL). Different from existing transfer learning problems which assume that there is a close relation between the source domain and the target domain, in the DDTL problem, the target domain can be totally different from the source domain. For example, the source domain classifies face images but the target domain distinguishes plane images. Inspired by the cognitive processof human where two seemingly unrelated concepts can be connected by learning intermediate concepts gradually, we propose a Selective Learning Algorithm (SLA) to solve the DDTL problem with supervised autoencoder or supervised convolutional autoencoder as a base model for handling different types of inputs. Intuitively, the SLA algorithm selects usefully unlabeled data gradually from intermediate domains as a bridge to break the large distribution gap for transferring knowledge between two distant domains. Empirical studies on image classification problems demonstrate the effectiveness of the proposed algorithm, and on some tasks the improvement in terms of the classification accuracy is up to 17% over “non-transfer” methods.

Convolutional Neural Networks (CNNs) have demonstrated state-of-the-art performance on many visual recognition tasks. However, the combination of convolution and pooling operations only shows invariance to small local location changes in meaningful objects in input. Sometimes, such networks are trained using data augmentation to encode this invariance into the parameters, which restricts the capacity of the model to learn the content of these objects. A more efficient use of the parameter budget is to encode rotation or translation invariance into the model architecture, which relieves the model from the need to learn them. To enable the model to focus on learning the content of objects other than their locations, we propose to conduct patch ranking of the feature maps before feeding them into the next layer. When patch ranking is combined with convolution and pooling operations, we obtain consistent representations despite the location of meaningful objects in input. We show that the patch ranking module improves the performance of the CNN on many benchmark tasks, including MNIST digit recognition, large-scale image recognition, and image retrieval.

In order to grapple with the conundrum in the scalability of kernel-based learning algorithms, the method of approximating nonlinear kernels via random feature maps has attracted wide attention in large-scale learning systems. Specifically, the associated sampling procedure is one critical component that dictates the quality of random feature maps. However, for high-dimensional features, the standard Monte Carlo sampling method has been shown to be less effective in producing low-variance random samples. In consequence, it demands constructing a large number of features to attain the desired accuracy for downstream use. In this paper, we present a novel sampling algorithm powered by moment matching techniques to reduce the variance of random features. Our extensive empirical studies and comparisons with several highly competitive peer methods verify the superiority of the proposed algorithm in Gram matrix approximation and generalization errors in regression. Our rigorous theoretical proofs justify that the proposed algorithm is guaranteed achieving lower variance than the standard Monte Carlo method in high dimensional settings.

We propose a fast first-order method to solve multi-term nonsmooth composite convex minimization problems by employing a recent proximal average approximation technique and a novel adaptive parameter tuning technique. Thanks to this powerful parameter tuning technique, the proximal gradient step can be performed with a much larger stepsize in the algorithm implementation compared with the prior PA-APG method, which is the core to enable significant improvements in practical performance. Moreover, by choosing the approximation parameter adaptively, the proposed method is shown to enjoy the O(1/k) iteration complexity theoretically without needing any extra computational cost, while the PA-APG method incurs much more iterations for convergence. The preliminary experimental results on overlapping group Lasso and graph-guided fused Lasso problems confirm our theoretic claim well, and indicate that the proposed method is almost five times faster than the state-of-the-art PA-APG method and therefore suitable for higher-precision required optimization.

In this paper, we recast the subspace clustering as a verification problem. Our idea comes from an assumption that the distribution between a given sample x and cluster centers Omega is invariant to different distance metrics on the manifold, where each distribution is defined as a probability map (i.e. soft-assignment) between x and Omega. To verify this so-called invariance of distribution, we propose a deep learning based subspace clustering method which simultaneously learns a compact representation using a neural network and a clustering assignment by minimizing the discrepancy between pair-wise sample-centers distributions. To the best of our knowledge, this is the first work to reformulate clustering as a verification problem. Moreover, the proposed method is also one of the first several cascade clustering models which jointly learn representation and clustering in end-to-end manner. Extensive experimental results show the effectiveness of our algorithm comparing with 11 state-of-the-art clustering approaches on four data sets regarding to four evaluation metrics.

A variety of feature selection methods based on sparsity regularization have been developed with different loss functions and sparse regularization functions. Capitalizing on the existing sparsity regularized feature selection methods, we propose a general sparsity feature selection (GSR-FS) algorithm that optimizes a ℓ 2, r (0 <  r ≤ 2) based loss function with a ℓ 2, p -norm (0 < p ≤ 2) sparse regularization. The ℓ 2, r - norm (0 < 𝑟 ≤ 2) based loss function brings flexibility to balance data-fitting and robustness to outliers by tuning its parameter, and the ℓ 2, p -norm (0 < p ≤ 1) based regularization function is able to boost the sparsity for feature selection. To solve the optimization problem with multiple non-smooth and non-convex functions when , we develop an efficient solver under the general umbrella of Iterative Reweighted Least Square (IRLS) algorithms. Our algorithm has been proved to converge with a theoretical convergence order of min(2 – r, 2 – p ) at least . The experimental results have demonstrated that our method could achieve competitive feature selection performance on publicly available datasets compared with state-of-the-art feature selection methods, with reduced computational cost.

This paper studies a top-k hierarchical classification problem. In top-k classification, one is allowed to make k predictions and no penalty is incurred if at least one of k predictions is correct. In hierarchical classification, classes form a structured hierarchy, and misclassification costs depend on the relation between the correct class and the incorrect class in the hierarchy. Despite that the fact that both top-k classification and hierarchical classification have gained increasing interests, the two problems have always been studied separately. In this paper, we define a top-k hierarchical loss function using a real world application. We provide the Bayes-optimal solution that minimizes the expected top-k hierarchical misclassification cost. Via numerical experiments, we show that our solution outperforms two baseline methods that address only one of the two issues.

Support vector machine (SVM) model is one of most successful machine learning methods and has been successfully applied to solve numerous real-world application. Because the SVM methods use the hinge loss or squared hinge loss functions for classifications, they usually outperform other classification approaches, e.g. the least square loss function based methods. However, like most supervised learning algorithms, they learn classifiers based on the labeled data in training set without specific strategy to deal with the noise data. In many real-world applications, we often have data outliers in train set, which could misguide the classifiers learning, such that the classification performance is suboptimal. To address this problem, we proposed a novel capped Lp-norm SVM classification model by utilizing the capped `p-norm based hinge loss in the objective which can deal with both light and heavy outliers. We utilize the new formulation to naturally build the multiclass capped Lp-norm SVM. More importantly, we derive a novel optimization algorithms to efficiently minimize the capped Lp-norm based objectives, and also rigorously prove the convergence of proposed algorithms. We present experimental results showing that employing the new capped Lp-norm SVM method can consistently improve the classification performance, especially in the cases when the data noise level increases.

Collective inference is widely used to improve classification in network datasets. However, despite recent advances in deep learning and the successes of recurrent neural networks (RNNs), researchers have only just recently begun to study how to apply RNNs to heterogeneous graph and network datasets. There has been recent work on using RNNs for unsupervised learning in networks (e.g., graph clustering, node embedding) and for prediction (e.g., link prediction, graph classification), but there has been little work on using RNNs for node-based relational classification tasks. In this paper, we provide an end-to-end learning framework using RNNs for collective inference. Our main insight is to transform a node and its set of neighbors into an unordered sequence (of varying length) and use an LSTM-based RNN to predict the class label as the output of that sequence. We develop a collective inference method, which we refer to as Deep Collective Inference (DCI), that uses semi-supervised learning in partially-labeled networks and two label distribution correction mechanisms for imbalanced classes. We compare to several alternative methods on seven network datasets. DCI achieves up to a 12% reduction in error compared to the best alternative and a 25% reduction in error on average — over all methods, for all label proportions.

Multivariate count data are pervasive in science in the form of histograms, contingency tables and others. Previous work on modeling this type of distributions do not allow for fast and tractable inference. In this paper we present a novel Poisson graphical model, the first based on sum product networks, called PSPN, allowing for positive as well as negative dependencies. We present algorithms for learning tree PSPNs from data as well as for tractable inference via symbolic evaluation. With these, information-theoretic measures such as entropy, mutual information, and distances among count variables can be computed without resorting to approximations. Additionally, we show a connection between PSPNs and LDA, linking the structure of tree PSPNs to a hierarchy of topics. The experimental results on several synthetic and real world datasets demonstrate that PSPN often outperform state-of-the-art while remaining tractable.