Feature noising is an effective mechanism on reducing the risk of overfitting. To avoid an explosive searching space, existing work typically assumes that all features share a single noise level, which is often cross-validated. In this paper, we present a Bayesian feature noising model that flexibly allows for dimension-specific or group-specific noise levels, and we derive a learning algorithm that adaptively updates these noise levels. Our adaptive rule is simple and interpretable, by drawing a direct connection to the fitness of each individual feature or feature group. Empirical results on various datasets demonstrate the effectiveness on avoiding extensive tuning and sometimes improving the performance due to its flexibility.

Research in learning from demonstration can generally be grouped into either imitation learning or intention learning. In imitation learning, the goal is to imitate the observed behavior of an expert and is typically achieved using supervised learning techniques. In intention learning, the goal is to learn the intention that motivated the expert's behavior and to use a planning algorithm to derive behavior. Imitation learning has the advantage of learning a direct mapping from states to actions, which bears a small computational cost. Intention learning has the advantage of behaving well in novel states, but may bear a large computational cost by relying on planning algorithms in complex tasks. In this work, we introduce receding horizon inverse reinforcement learning, in which the planning horizon induces a continuum between these two learning paradigms. We present empirical results on multiple domains that demonstrate that performing IRL with a small, but non-zero, receding planning horizon greatly decreases the computational cost of planning while maintaining superior generalization performance compared to imitation learning.

Random feature mappings have been successfully used for approximating non-linear kernels to scaleup kernel methods. Some work aims at speeding up the feature mappings, but brings increasing variance of the approximation. In this paper, we propose a novel random feature mapping method that uses a signed Circulant Random Matrix (CRM) instead of an unstructured random matrix to project input data. The signed CRM has linear space complexity as the whole signed CRM can be recovered from one column of the CRM, and ensures loglin-ear time complexity to compute the feature mapping using the Fast Fourier Transform (FFT). Theoretically, we prove that approximating Gaussian kernel using our mapping method is unbiased and does not increase the variance. Experimentally, we demonstrate that our proposed mapping method is time and space efficient while retaining similar accuracies with state-of-the-art random feature mapping methods. Our proposed random feature mapping method can be implemented easily and make kernel methods scalable and practical for large scale training and predicting problems.

The rapidly growing of Location-based Social Networks (LBSNs) provides a vast amount of check-in data, which enables many services, e.g., point-of-interest (POI) recommendation. In this paper, we study the next new POI recommendation problem in which new POIs with respect to users' current location are to be recommended. The challenge lies in the difficulty in precisely learning users' sequential information and personalizing the recommendation model. To this end, we resort to the Metric Embedding method for the recommendation, which avoids drawbacks of the Matrix Factorization technique. We propose a personalized ranking metric embedding method (PRME) to model personalized check-in sequences. We further develop a PRME-G model, which integrates sequential information, individual preference, and geographical influence, to improve the recommendation performance. Experiments on two real-world LBSN datasets demonstrate that our new algorithm outperforms the state-of-the-art next POI recommendation methods.

We address an ensemble clustering problem, where reliable clusters are locally embedded in given multiple partitions. We propose a new nonnegative matrix factorization (NMF)-based method, in which locally reliable clusters are explicitly considered by using instance-wise weights over clusters. Our method factorizes the input cluster assignment matrix into two matrices H and W, which are optimized by iteratively 1) updating H and W while keeping the weight matrix constant and 2) updating the weight matrix while keeping H and W constant, alternatively. The weights in the second step were updated by solving a convex problem, which makes our algorithm significantly faster than existing NMF-based ensemble clustering methods. We empirically proved that our method outperformed a lot of cutting-edge ensemble clustering methods by using a variety of datasets.

The rapidly growing of Location-based Social Networks (LBSNs) provides a vast amount of check-in data, which enables many services, e.g., point-of-interest (POI) recommendation. In this paper, we study the next new POI recommendation problem in which new POIs with respect to users' current location are to be recommended. The challenge lies in the difficulty in precisely learning users' sequential information and personalizing the recommendation model. To this end, we resort to the Metric Embedding method for the recommendation, which avoids drawbacks of the Matrix Factorization technique. We propose a personalized ranking metric embedding method (PRME) to model personalized check-in sequences. We further develop a PRME-G model, which integrates sequential information, individual preference, and geographical influence, to improve the recommendation performance. Experiments on two real-world LBSN datasets demonstrate that our new algorithm outperforms the state-of-the-art next POI recommendation methods.

In this paper we investigate the use of coactive learning in a multitask setting. In coactive learning, an expert presents the learner with a problem and the learner returns a candidate solution. The expert then improves on the solution if necessary and presents the improved solution to the learner. The goal for the learner is to learn to produce solutions which cannot be further improved by the expert while minimizing the average expert effort. In this paper, we consider the setting where there are multiple experts (tasks), and in each iteration one expert presents a problem to the learner. While the experts are expected to have different solution preferences, they are also assumed to share similarities, which should enable generalization across experts. We analyze several algorithms for this setting and derive bounds on the average expert effort during learning. Our main contribution is the balanced Perceptron algorithm, which is the first coactive learning algorithm that is both able to generalize across experts when possible, while also guaranteeing convergence to optimal solutions for individual experts. Our experiments in three domains confirm that this algorithm is effective in the multitask setting, compared to natural baselines.

Principal component analysis (PCA) is an unsupervised method for learning low-dimensional features with orthogonal projections. Multilinear PCA methods extend PCA to deal with multidimensional data (tensors) directly via tensor-to-tensor projection or tensor-to-vector projection (TVP). However, under the TVP setting, it is difficult to develop an effective multilinear PCA method with the orthogonality constraint. This paper tackles this problem by proposing a novel Semi-Orthogonal Multilinear PCA (SO-MPCA) approach. SO-MPCA learns low-dimensional features directly from tensors via TVP by imposing the orthogonality constraint in only one mode. This formulation results in more captured variance and more learned features than full orthogonality. For better generalization, we further introduce a relaxed start (RS) strategy to get SO-MPCA-RS by fixing the starting projection vectors, which increases the bias and reduces the variance of the learning model. Experiments on both face (2D) and gait (3D) data demonstrate that SO-MPCA-RS outperforms other competing algorithms on the whole, and the relaxed start strategy is also effective for other TVP-based PCA methods.

Gaussian processes (GPs) provide a nonparametric representation of functions. However, classical GP inference suffers from high computational cost for big data. In this paper, we propose a new Bayesian approach, EigenGP, that learns both basis dictionary elements — eigenfunctions of a GP prior — and prior precisions in a sparse finite model. It is well known that, among all orthogonal basis functions, eigenfunctions can provide the most compact representation. Unlike other sparse Bayesian finite models where the basis function has a fixed form, our eigenfunctions live in a reproducing kernel Hilbert space as a finite linear combination of kernel functions. We learn the dictionary elements — eigenfunctions — and the prior precisions over these elements as well as all the other hyperparameters from data by maximizing the model marginal likelihood. We explore computational linear algebra to simplify the gradient computation significantly. Our experimental results demonstrate improved predictive performance of EigenGP over alternative sparse GP methods as well as relevance vector machines.

Large sparse datasets are common in many real-world applications. Linear SVM has been shown to be very efficient for classifying such datasets. However, it is still unknown how data sparseness would affect its convergence behavior. To study this problem in a systematic manner, we propose a novel approach to generate large and sparse data from real-world datasets, using statistical inference and the data sampling process in the PAC framework. We first study the convergence behavior of linear SVM experimentally, and make several observations, useful for real-world applications. We then offer theoretical proofs for our observations by studying the Bayes risk and PAC bound. Our experiment and theoretic results are valuable for learning large sparse datasets with linear SVM.