Search

Accurate prediction of individuals’ brain age is critical to establish a baseline for normal brain development. This study proposes to model brain development with a novel non-negative projective dictionary learning (NPDL) approach, which learns a discriminative representation of multi-modal neuroimaging data for predicting brain age. Our approach encodes the variability of subjects in different age groups using separate dictionaries, projecting features into a low-dimensional manifold such that information is preserved only for the corresponding age group. The proposed framework improves upon previous discriminative dictionary learning methods by inco

Multi-label classification is a central problem in many application domains. In this paper, we present a novel supervised bi-directional model that learns a low-dimensional mid-level representation for multi-label classification. Unlike traditional multi-label learning methods which identify intermediate representations from either the input space or the output space but not both, the mid-level representation in our model has two complementary parts that capture intrinsic information of the input data and the output labels respectively under the autoencoder principle while augmenting each other for the target output label prediction. The resulting optimization problem can be solved efficiently using an iterative procedure with alternating steps, while closed-form solutions exist for one major step. Our experiments conducted on a variety of multi-label data sets demonstrate the efficacy of the proposed bi-directional representation learning model for multi-label classification.

Semantic scene classification is a challenging problem in computer vision. In this paper, we present a novel multi-level active learning approach to reduce the human annotation effort for training robust scene classification models. Different from most existing active learning methods that can only query labels for selected instances at the target categorization level, i.e., the scene class level, our approach establishes a semantic framework that predicts scene labels based on a latent object-based semantic representation of images, and is capable to query labels at two different levels, the target scene class level (abstractive high level) and the latent object class level (semantic middle level). Specifically, we develop an adaptive active learning strategy to perform multi-level label query, which maintains the default label query at the target scene class level, but switches to the latent object class level whenever an "unexpected" target class label is returned by the labeler. We conduct experiments on two standard scene classification datasets to investigate the efficacy of the proposed approach. Our empirical results show the proposed adaptive multi-level active learning approach can outperform both baseline active learning methods and a state-of-the-art multi-level active learning method.

Although multi-label classification has become an increasingly important problem in machine learning, current approaches remain restricted to learning in the original label space (or in a simple linear projection of the original label space). Instead, we propose to use kernels on output label vectors to significantly expand the forms of label dependence that can be captured. The main challenge is to reformulate standard multi-label losses to handle kernels between output vectors. We first demonstrate how a state-of-the-art large margin loss for multi-label classification can be reformulated, exactly, to handle output kernels as well as input kernels. Importantly, the pre-image problem for multi-label classification can be easily solved at test time, while the training procedure can still be simply expressed as a quadratic program in a dual parameter space. We then develop a projected gradient descent training procedure for this new formulation. Our empirical results demonstrate the efficacy of the proposed approach on complex image labeling tasks.

Labeled data is often sparse in common learning scenarios, either because it is too time consuming or too expensive to obtain, while unlabeled data is almost always plentiful. This asymmetry is exacerbated in multi-label learning, where the labeling process is more complex than in the single label case. Although it is important to consider semi-supervised methods for multi-label learning, as it is in other learning scenarios, surprisingly, few proposals have been investigated for this particular problem. In this paper, we present a new semi-supervised multi-label learning method that combines large-margin multi-label classification with unsupervised subspace learning. We propose an algorithm that learns a subspace representation of the labeled and unlabeled inputs, while simultaneously training a supervised large-margin multi-label classifier on the labeled portion. Although joint training of these two interacting components might appear intractable, we exploit recent developments in induced matrix norm optimization to show that these two problems can be solved jointly, globally and efficiently. In particular, we develop an efficient training procedure based on subgradient search and a simple coordinate descent strategy. An experimental evaluation demonstrates that semi-supervised subspace learning can improve the performance of corresponding supervised multi-label learning methods.

The analysis of social networks often assumes time invariant scenario, while in practice actor attributes and links in such networks often evolve over time and are inextricably dependent on each other. In this article, we propose a new method to predict actor attributes and links in temporal networks. Our approach takes into account the attributes corresponding to the participating actors together with topological and structural changes of the network over time. This is achieved by building two conditional predictors to jointly infer links and actor attributes. The proposed prediction method was significantly more accurate than alternatives when evaluated on synthetic data sets and two well-studied real-life temporal social networks. In addition, the new algorithm is computationally more efficient than a related alternative scaling up linearly with the number of temporal observations and quadratically with the number of actors considered.

We present a generalized view of support vector machines that does not rely on a Euclidean geometric interpretation nor even positive semidefinite kernels. We base our development instead on the confidence matrix-the matrix normally determined by the direct (Hadamard) product of the kernel matrix with the label outer-product matrix. It turns out that alternative forms of confidence matrices are possible, and indeed useful. By focusing on the confidence matrix instead of the underlying kernel, we can derive an intuitive principle for optimizing example weights to yield robust classifiers. Our principle initially recovers the standard quadratic SVM training criterion, which is only convex for kernel-derived confidence measures. However, given our generalized view, we are then able to derive a principled relaxation of the SVM criterion that yields a convex upper bound. This relaxation is always convex and can be solved with a linear program. Our new training procedure obtains similar generalization performance to standard SVMs on kernel-derived confidence functions, but achieves even better results with indefinite confidence functions.

In this paper, we present a novel semidefinite programming approach for multiple-instance learning. We first formulate the multiple-instance learning as a combinatorial maximum margin optimization problem with additional instance selection constraints within the framework of support vector machines. Although solving this primal problem requires non-convex programming, we nevertheless can then derive an equivalent dual formulation that can be relaxed into a novel convex semidefinite programming (SDP). The relaxed SDP has free parameters where T is the number of instances, and can be solved using a standard interior-point method. Empirical study shows promising performance of the proposed SDP in comparison with the support vector machine approaches with heuristic optimization procedures.