Accelerating difficulty estimation for conformal regression forests

The conformal prediction framework allows for specifying the probability of making incorrect predictions by a user-provided confidence level. In addition to a learning algorithm, the framework requires a real-valued function, called nonconformity measure, to be specified. The nonconformity measure does not affect the error rate, but the resulting efficiency, i.e., the size of output prediction regions, may vary substantially. A recent large-scale empirical evaluation of conformal regression approaches showed that using random forests as the learning algorithm together with a nonconformity measure based on out-of-bag errors normalized using a nearest-neighbor-based difficulty estimate, resulted in state-of-the-art performance with respect to efficiency. However, the nearest-neighbor procedure incurs a significant computational cost. In this study, a more straightforward nonconformity measure is investigated, where the difficulty estimate employed for normalization is based on the variance of the predictions made by the trees in a forest. A large-scale empirical evaluation is presented, showing that both the nearest-neighbor-based and the variance-based measures significantly outperform a standard (non-normalized) nonconformity measure, while no significant difference in efficiency between the two normalized approaches is observed. The evaluation moreover shows that the computational cost of the variance-based measure is several orders of magnitude lower than when employing the nearest-neighbor-based nonconformity measure. The use of out-of-bag instances for calibration does, however, result in nonconformity scores that are distributed differently from those obtained from test instances, questioning the validity of the approach. An adjustment of the variance-based measure is presented, which is shown to be valid and also to have a significant positive effect on the efficiency. For conformal regression forests, the variance-based nonconformity measure is hence a computationally efficient and theoretically well-founded alternative to the nearest-neighbor procedure.     

Generalized random shapelet forests

Shapelets are discriminative subsequences of time series, usually embedded in shapelet-based decision trees. The enumeration of time series shapelets is, however, computationally costly, which in addition to the inherent difficulty of the decision tree learning algorithm to effectively handle high-dimensional data, severely limits the applicability of shapelet-based decision tree learning from large (multivariate) time series databases. This paper introduces a novel tree-based ensemble method for univariate and multivariate time series classification using shapelets, called the generalized random shapelet forest algorithm. The algorithm generates a set of shapelet-based decision trees, where both the choice of instances used for building a tree and the choice of shapelets are randomized. For univariate time series, it is demonstrated through an extensive empirical investigation that the proposed algorithm yields predictive performance comparable to the current state-of-the-art and significantly outperforms several alternative algorithms, while being at least an order of magnitude faster. Similarly for multivariate time series, it is shown that the algorithm is significantly less computationally costly and more accurate than the current state-of-the-art.     

Cost-efficient drilling using industrial robots with high-bandwidth force feedback

Here we present a method for high-precision drilling using an industrial robot with high-bandwidth force feedback, which is used for building up pressure to clamp-up an end-effector to the work-piece surface prior to drilling. The focus is to eliminate the sliding movement (skating) of the end-effector during the clamp-up of the end-effector to the work-piece surface, an undesired effect that is due to the comparatively low mechanical stiffness of typical serial industrial robots. This compliance also makes the robot deflect due to the cutting forces, resulting in poor hole position accuracy and to some extent in poor hole quality. Recently, functionality for high-bandwidth force control has found its way into industrial robot control systems. This could potentially open up the possibility for robotic drilling systems with improved performance, using only standard systems without excessive extra hardware and calibration techniques. Instead of automation with expensive fixtures and precise machinery, our approach was to make use of standard low-cost robot equipment in combination with sensor feedback. The resulting sliding suppression control results in greatly improved hole positioning and quality. The conceptual idea behind the force control is useful also in many other robotic applications requiring external sensor feedback control.     

6D scan registration using depth-interpolated local image features

This paper describes a novel registration approach that is based on a combination of visual and 3D range information. To identify correspondences, local visual features obtained from images of a standard color camera are compared and the depth of matching features (and their position covariance) is determined from the range measurements of a 3D laser scanner. The matched depth-interpolated image features allow one to apply registration with known correspondences. We compare several ICP variants in this paper and suggest an extension that considers the spatial distance between matching features to eliminate false correspondences. Experimental results are presented in both outdoor and indoor environments. In addition to pair-wise registration, we also propose a global registration method that registers all scan poses simultaneously.     

A layout technique for class diagrams to be used in product configuration projects

The use of product configurators has produced a range of benefits for several companies, such as minimizing the use of resources and shortening the lead times in product specification processes. When developing a product configurator, two kinds of models are often created, namely analysis models and design models. Since the task of describing product knowledge in analysis models involves domain experts, the analysis language has to be easily understandable in order to avoid the need for extensive training. For this task, the so-called Product Variant Master (PVM) diagramming technique is often applied. With regard to the design model, the requirements for the language focus more on a formalized and rich language, which is why class diagrams are often applied. To avoid the use of different modelling languages in the analysis and design phase, this paper proposes the layout technique ‘Vertically Aligned Class Diagrams’ (VACDs), which incorporate the usability of PVMs into class diagrams. To validate the usefulness of the VACD technique, the paper compares VACDs to PVMs and class diagrams in a utility analysis and a usability experiment. These investigations strongly indicate that VACDs maintain to a great extent the combined advantages of PVMs and normally drawn class diagrams. Thus, the use of VACDs in configurator projects has the potential to increase efficiency, improve communication and reduce errors.     

Time-adaptive quantile regression

An algorithm for time-adaptive quantile regression is presented. The algorithm is based on the simplex algorithm, and the linear optimization formulation of the quantile regression problem is given. The observations have been split to allow a direct use of the simplex algorithm. The simplex method and an updating procedure are combined into a new algorithm for time-adaptive quantile regression, which generates new solutions on the basis of the old solution, leading to savings in computation time. The suggested algorithm is tested against a static quantile regression model on a data set with wind power production, where the models combine splines and quantile regression. The comparison indicates superior performance for the time-adaptive quantile regression in all the performance parameters considered.¹     

Data-Mining-Cup 2007

Zusammenfassung  Die Aufgabe des Data-Mining-Cups 2007 war die Entwicklung einer vollautomatischen Methode zur effektiven Steuerung des Einsatzes von Rabattcoupons. Vorverarbeitung und Parameter wurden durch eine Kreuzvalidierung optimiert. Experimente zeigen, dass ein einzelner Klassifikator in der Regel nicht genügt, die Aufgabenstellung mit ausreichender Güte zu l?sen. Vielmehr müssen mehrere, m?glichst verschiedene, Klassifikatoren dazu herangezogen werden. Die eingereichten L?sungen entstanden durch Kombination von bis zu 2.000 Klassifikatoren und haben im Wettbewerb sehr gut abgeschnitten: Sechs von neun Abgaben lagen unter den ersten zehn Pl?tzen, die übrigen drei L?sungen sind immer noch unter den ersten 20 von insgesamt 230 Pl?tzen.     

Organizational Adaptation in Multi‐Stakeholder Crisis Response: An Experimental Study

Modern day crises demand organizations to collaborate and adapt to new roles, functions and structures. In such situations, lack of collaborative behaviour and openness between organizations can result in reduced adaptive ability. Therefore, it is important to facilitate collaboration between organizations. We have studied the extent to which crisis managers are prepared to work with personnel and resources from organizations other than their own when responding to crises. An experiment was designed with four different organizations in Sweden, which involved decision making concerning whether the participants systematically favoured their own organization over others. Findings indicate that increasing familiarity and expectation of future cooperation with other organizations increased the likelihood that decision makers would be prepared to work with other organizations in joint crisis management.     

Textures and spin waves in rotating superfluid3H3-B

The spin wave modes in rotating superfluid³He-B have been calculated for a texture of cylindrical symmetry. The textural configuration is determined by minimizing the appropriate free energy and solving the resulting Euler-Lagrange equations for then vector. The texture determines the effective potential entering the spin wave equation. The potential deviates significantly from a harmonic oscillator under realistic experimental conditions, giving rise to shifts of the oscillator frequencies. The agreement with the recent experimental observations of Ikkalaet al. is excellent at all temperatures studied.