Methods for combining predictions from different models in a supervised learning setting must somehow estimate/predict the quality of a model's predictions at unknown future inputs. Many of these methods (often implicitly) make the assumption that the test inputs are identical to the training inputs, which is seldom reasonable. By failing to take into account that prediction will generally be harder for test inputs that did not occur in the training set, this leads to the selection of too complex models. Based on a novel, unbiased expression for KL divergence, we propose XAIC and its special case FAIC as versions of AIC intended for prediction that use different degrees of knowledge of the test inputs. Both methods substantially differ from and may outperform all the known versions of AIC even when the training and test inputs are iid, and are especially useful for deterministic inputs and under covariate shift. Our experiments on linear models suggest that if the test and training inputs differ substantially, then XAIC and FAIC predictively outperform AIC, BIC and several other methods including Bayesian model averaging.

Noname manuscript No. (will be inserted by the editor) Abstract Many applications in data analysis begin with a set of points in a Euclidean space that is partitioned into clusters. Common tasks then are to devise a classifier deciding which of the clusters a new point is associated to, finding outliers with respect to the clusters, or identifying the type of clustering used for the partition. One of the common kinds of clusterings are (balanced) least-squares assignments with respect to a given set of sites. For these, there is a'separating power diagram' for which each cluster lies in its own cell. In the present paper, we aim for efficient algorithms for outlier detection and the computation of thresholds that measure how similar a clustering is to a leastsquares assignment for fixed sites. For this purpose, we devise a new model for the computation of a'soft power diagram', which allows a soft separation of the clusters with'point counting properties'; e.g. As our results hold for a more general non-convex model of free sites, we describe it and our proofs in this more general way. Its locally optimal solutions satisfy the aforementioned point counting properties. For our target applications that use fixed sites, our algorithms are efficiently solvable to global optimality by linear programming.

When interacting with a robot, users often under-specify the tasks to be performed. For example in Figure 5, when asked to pour something, the robot has to infer which cup to pour into and a complete sequence of the navigation and manipulation steps--moving close, grasping, placing, and so on. This sequence not only changes with the task, but also with the perceived state of the environment. As an example, consider the task of a robot fetching a magazine from a desk. The method to perform this task varies depending on several properties of the environment: for example, the robot's relative distance from the magazine, the robot's relative orientation, the thickness of the magazine, and the presence or the absence of other items on top of the magazine. If the magazine is very thin, the robot may have to slide the magazine to the side of the table to pick it up. If there is a mug sitting on top of the magazine, it would have to be moved prior to the magazine being picked up. Thus, especially when the details of the manipulation task are under-specified, the success of executing the task depends on the ability to detect the object and on the ability to sequence the set of primitives (navigation and manipulation controllers) in various ways in response to the environment. In recent years, there have been significant developments in building low-level controllers for robots [34] as well as in perceptual tasks such as object detection from sensor data [20, 11, 35].

We study the basic problem of robust subspace recovery. That is, we assume a data set that some of its points are sampled around a fixed subspace and the rest of them are spread in the whole ambient space, and we aim to recover the fixed underlying subspace. We first estimate "robust inverse sample covariance" by solving a convex minimization procedure; we then recover the subspace by the bottom eigenvectors of this matrix (their number correspond to the number of eigenvalues close to 0). We guarantee exact subspace recovery under some conditions on the underlying data. Furthermore, we propose a fast iterative algorithm, which linearly converges to the matrix minimizing the convex problem. We also quantify the effect of noise and regularization and discuss many other practical and theoretical issues for improving the subspace recovery in various settings. When replacing the sum of terms in the convex energy function (that we minimize) with the sum of squares of terms, we obtain that the new minimizer is a scaled version of the inverse sample covariance (when exists). We thus interpret our minimizer and its subspace (spanned by its bottom eigenvectors) as robust versions of the empirical inverse covariance and the PCA subspace respectively. We compare our method with many other algorithms for robust PCA on synthetic and real data sets and demonstrate state-of-the-art speed and accuracy.

Recent work has demonstrated that problems-- particularly imitation learning and structured prediction-- where a learner's predictions influence the input-distribution it is tested on can be naturally addressed by an interactive approach and analyzed using no-regret online learning. These approaches to imitation learning, however, neither require nor benefit from information about the cost of actions. We extend existing results in two directions: first, we develop an interactive imitation learning approach that leverages cost information; second, we extend the technique to address reinforcement learning. The results provide theoretical support to the commonly observed successes of online approximate policy iteration. Our approach suggests a broad new family of algorithms and provides a unifying view of existing techniques for imitation and reinforcement learning.

Software design is crucial to successful software development, yet is a demanding multi-objective problem for software engineers. In an attempt to assist the software designer, interactive (i.e. human in-the-loop) meta-heuristic search techniques such as evolutionary computing have been applied and show promising results. Recent investigations have also shown that Ant Colony Optimization (ACO) can outperform evolutionary computing as a potential search engine for interactive software design. With a limited computational budget, ACO produces superior candidate design solutions in a smaller number of iterations. Building on these findings, we propose a novel interactive ACO (iACO) approach to assist the designer in early lifecycle software design, in which the search is steered jointly by subjective designer evaluation as well as machine fitness functions relating the structural integrity and surrogate elegance of software designs. Results show that iACO is speedy, responsive and highly effective in enabling interactive, dynamic multi-objective search in early lifecycle software design. Study participants rate the iACO search experience as compelling. Results of machine learning of fitness measure weightings indicate that software design elegance does indeed play a significant role in designer evaluation of candidate software design. We conclude that the evenness of the number of attributes and methods among classes (NAC) is a significant surrogate elegance measure, which in turn suggests that this evenness of distribution, when combined with structural integrity, is an implicit but crucial component of effective early lifecycle software design.

This paper proposes a model, the linear model, for randomly generating logic programs with low density of rules and investigates statistical properties of such random logic programs. It is mathematically shown that the average number of answer sets for a random program converges to a constant when the number of atoms approaches infinity. Several experimental results are also reported, which justify the suitability of the linear model. It is also experimentally shown that, under this model, the size distribution of answer sets for random programs tends to a normal distribution when the number of atoms is sufficiently large. KEYWORDS: answer set programming, random logic programs.

We reduce a broad class of machine learning problems, usually addressed by EM or sampling, to the problem of finding the $k$ extremal rays spanning the conical hull of a data point set. These $k$ "anchors" lead to a global solution and a more interpretable model that can even outperform EM and sampling on generalization error. To find the $k$ anchors, we propose a novel divide-and-conquer learning scheme "DCA" that distributes the problem to $\mathcal O(k\log k)$ same-type sub-problems on different low-D random hyperplanes, each can be solved by any solver. For the 2D sub-problem, we present a non-iterative solver that only needs to compute an array of cosine values and its max/min entries. DCA also provides a faster subroutine for other methods to check whether a point is covered in a conical hull, which improves algorithm design in multiple dimensions and brings significant speedup to learning. We apply our method to GMM, HMM, LDA, NMF and subspace clustering, then show its competitive performance and scalability over other methods on rich datasets.

Classical multidimensional scaling only works well when the noisy distances observed in a high dimensional space can be faithfully represented by Euclidean distances in a low dimensional space. Advanced models such as Maximum Variance Unfolding (MVU) and Minimum Volume Embedding (MVE) use Semi-Definite Programming (SDP) to reconstruct such faithful representations. While those SDP models are capable of producing high quality configuration numerically, they suffer two major drawbacks. One is that there exist no theoretically guaranteed bounds on the quality of the configuration. The other is that they are slow in computation when the data points are beyond moderate size. In this paper, we propose a convex optimization model of Euclidean distance matrices. We establish a non-asymptotic error bound for the random graph model with sub-Gaussian noise, and prove that our model produces a matrix estimator of high accuracy when the order of the uniform sample size is roughly the degree of freedom of a low-rank matrix up to a logarithmic factor. Our results partially explain why MVU and MVE often work well. Moreover, we develop a fast inexact accelerated proximal gradient method. Numerical experiments show that the model can produce configurations of high quality on large data points that the SDP approach would struggle to cope with.

Finding the optimal $k$-means clustering is NP-hard in general and many heuristics have been designed for minimizing monotonically the $k$-means objective. We first show how to extend Lloyd's batched relocation heuristic and Hartigan's single-point relocation heuristic to take into account empty-cluster and single-point cluster events, respectively. Those events tend to increasingly occur when $k$ or $d$ increases, or when performing several restarts. First, we show that those special events are a blessing because they allow to partially re-seed some cluster centers while further minimizing the $k$-means objective function. Second, we describe a novel heuristic, merge-and-split $k$-means, that consists in merging two clusters and splitting this merged cluster again with two new centers provided it improves the $k$-means objective. This novel heuristic can improve Hartigan's $k$-means when it has converged to a local minimum. We show empirically that this merge-and-split $k$-means improves over the Hartigan's heuristic which is the {\em de facto} method of choice. Finally, we propose the $(k,l)$-means objective that generalizes the $k$-means objective by associating the data points to their $l$ closest cluster centers, and show how to either directly convert or iteratively relax the $(k,l)$-means into a $k$-means in order to reach better local minima.