Real-world relational data are seldom stationary, yet traditional collaborative filtering algorithms generally rely on this assumption. Motivated by our sales prediction problem, we propose a factor-based algorithm that is able to take time into account. By introducing additional factors for time, we formalize this problem as a tensor factorization with a special constraint on the time dimension. Further, we provide a fully Bayesian treatment to avoid tuning parameters and achieve automatic model complexity control. To learn the model we develop an efficient sampling procedure that is capable of analyzing large-scale data sets. This new algorithm, called Bayesian Probabilistic Tensor Factorization (BPTF), is evaluated on several real-world problems including sales prediction and movie recommendation. Empirical results demonstrate the superiority of our temporal model.

Autonomous driving presents one of the largest problems that the robotics and artificial intelligence communities are facing at the moment, both in terms of difficulty and potential societal impact. Self-driving vehicles (SDVs) are expected to prevent road accidents and save millions of lives while improving the livelihood and life quality of many more. However, despite large interest and a number of industry players working in the autonomous domain, there still remains more to be done in order to develop a system capable of operating at a level comparable to best human drivers. One reason for this is high uncertainty of traffic behavior and large number of situations that an SDV may encounter on the roads, making it very difficult to create a fully generalizable system. To ensure safe and efficient operations, an autonomous vehicle is required to account for this uncertainty and to anticipate a multitude of possible behaviors of traffic actors in its surrounding. We address this critical problem and present a method to predict multiple possible trajectories of actors while also estimating their probabilities. The method encodes each actor's surrounding context into a raster image, used as input by deep convolutional networks to automatically derive relevant features for the task. Following extensive offline evaluation and comparison to state-of-the-art baselines, the method was successfully tested on SDVs in closed-course tests.

Abstract| Many control problems in the robotics eld can be cast as Partially Observed Markovian Decision Problems POMDPs, an optimal control formalism. Finding optimal solutions to such problems in general, however is known to be intractable. It has often been observed that in practice, simple structured controllers su ce for good sub-optimal control, and recent research in the arti cial intelligence community has focused on policy search methods as techniques for nding sub-optimal controllers when such structured controllers do exist. Traditional model-based reinforcement learning algorithms make a certainty equivalence assumption on their learned models and calculate optimal policies for a maximumlikelihood Markovian model. In this work, we consider algorithms that evaluate and synthesize controllers under distributions of Markovian models. Previous work has demonstrated that algorithms that maximize mean reward with respect to model uncertainty leads to safer and more robust controllers. We consider brie y other performance criterion that emphasize robustness and exploration in the search for controllers, and note the relation with experiment design and active learning. To v alidate the power of the approach on a robotic application we demonstrate the presented learning control algorithm by ying an autonomous helicopter. We show that the controller learned is robust and delivers good performance in this real-world domain.

No abstract

The false-discovery rate (FDR) is a new statistical procedure to control the number of mistakes made when performing multiple hypothesis tests, i.e., when comparing many data against a given model hypothesis. The key advantage of FDR is that it allows one to a priori control the average fraction of false rejections made (when comparing with the null hypothesis) over the total number of rejections performed. We compare FDR with the standard procedure of rejecting all tests that do not match the null hypothesis above some arbitrarily chosen conÐdence limit, e.g., 2 p, or at the 95% conÐdence level. We Ðnd a similar rate of correct detections, but with signiÐcantly fewer false detections. Moreover, the FDR procedure is quick and easy to compute and can be trivially adapted to work with correlated data. The purpose of this paper is to introduce the FDR procedure to the astrophysics community. We illustrate the power of FDR through several astronomical examples, including the detection of features against a smooth one-dimensional function, e.g., seeing the "" baryon wiggles ÏÏ in a power spectrum of matter Ñuc-tuations, and source pixel detection in imaging data. In this era of large data sets and high-precision measurements, FDR provides the means to adaptively control a scientiÐcally meaningful quantityÈthe fraction of false discoveries over total discoveries.

No abstract

We present measurements of the normalized redshift-space three-point correlation function (3PCF) (Q z ) of galaxies from the Sloan Digital Sky Survey (SDSS) main galaxy sample. These measurements were possible because of a fast new N-point correlation function algorithm (called npt) based on multiresolutional k-d trees. We have applied npt to both a volume-limited (36 738 galaxies with 0.05 z 0.095 and −23 M 0.0 r −20.5) and magnitude-limited sample (134 741 galaxies over 0.05 z 0.17 and ∼M * ± 1.5) of SDSS galaxies, and find consistent results between the two samples, thus confirming the weak luminosity dependence of Q z recently seen by other authors. We compare our results to other Q z measurements in the literature and find it to be consistent within the full jackknife error estimates. However, we find these errors are significantly increased by the presence of the 'Sloan Great Wall' (at z ∼0.08) within these two SDSS data sets, which changes the 3PCF by 70 per cent on large scales (s 10 h −1 Mpc). If we exclude this supercluster, our observed Q z is in better agreement with that obtained from the 2-degree Field Galaxy Redshift Survey (2dFGRS) by other authors, thus demonstrating the sensitivity of these higher order correlation functions to large-scale structures in the Universe. This analysis highlights that the SDSS data sets used here are not 'fair samples' of the Universe for the estimation of higher order clustering statistics and larger volumes are required. We study the shape dependence of Q z (s, q, θ) as one expects this measurement to depend on scale if the large-scale structure in the Universe has grown

We consider the problem of detecting anomalies in high arity categorical datasets. In most applications, anomalies are defined as data points that are 'abnormal'. Quite often we have access to data which consists mostly of normal records, along with a small percentage of unlabelled anomalous records. We are interested in the problem of unsupervised anomaly detection, where we use the unlabelled data for training, and detect records that do not follow the definition of normality.A standard approach is to create a model of normal data, and compare test records against it. A probabilistic approach builds a likelihood model from the training data. Records are tested for anomalousness based on the complete record likelihood given the probability model. For categorical attributes, bayes nets give a standard representation of the likelihood. While this approach is good at finding outliers in the dataset, it often tends to detect records with attribute values that are rare. Sometimes, just detecting rare values of an attribute is not desired and such outliers are not considered as anomalies in that context. We present an alternative definition of anomalies, and propose an approach of comparing against marginal distributions of attribute subsets. We show that this is a more meaningful way of detecting anomalies, and has a better performance over semi-synthetic as well as real world datasets.