Raven's Progressive Matrices and similar matrix problems have been used in research and intelligence testing for decades. The matrix problems serve as a nonverbal test of analogical reasoning and are thought to measure analytical intelligence (Carpenter, Just, & Shell, 1990), also known as fluid intelligence (cf. Cattell, 1963). Although these matrix problems have a wide variety of research applications, the relatively small number of matrices in Raven's original sets (108 total;Raven, Court, & Raven, 1998) limits their utility in several domains, including neuroimaging experiments and computational modeling of cognitive processes.Our goal in the present study was to create and characterize a very large set of matrix problems that have properties similar to those of Raven's original matrices. We sought to create the matrix set in a systematic way that would allow researchers to have a great deal of control over the underlying structure, surface features, and difficulty of the matrix problems. This in turn would allow researchers to systematically expand the range of difficulty in their stimulus sets beyond the range provided by the original Raven's matrices. To accomplish these goals, we analyzed the underlying structures in Raven's original Standard Progressive Matrices (SPMs) to determine what types and combinations of relations were used. On the basis of that analysis, we developed software that can use the same underlying patterns to generate large numbers of unique matrix problems using parameters chosen by the researcher. Specifically, the software is designed so that researchers can choose the type, direction, and number of relations in a problem and create any number of unique matrices that share the same underlying structure (e.g., changes in numerosity in a diagonal pattern) but have different surface features (e.g., shapes, colors).Finally, we used the matrix generation software to produce a representative set of matrix problems that cover the range of underlying structures that can be produced by the software. This set of matrices was compared with Raven's SPMs in a norming study. The first goal of the norming study was to compare the difficulty of the generated matrices with the difficulty of the SPMs with the same underlying structure. The second goal was to assess the difficulty of specific structural features within the matrices and the range of problem difficulties that can be produced by the matrix generation software when those features are combined. Analysis of Raven's Progressive Matrix StructuresPrevious studies have analyzed the factors that contribute to the difficulty of Raven and Raven-like matrix problems. Raven's Progressive Matrices is a widely used test for assessing intelligence and reasoning ability (Raven, Court, & Raven, 1998). Since the test is nonverbal, it can be applied to many different populations and has been used all over the world (Court & Raven, 1995). However, relatively few matrices are in the sets developed by Raven, which limits their use in experiments requiring l...

COMET is a single-pass MapReduce algorithm for learning on large-scale data. It builds multiple random forest ensembles on distributed blocks of data and merges them into a mega-ensemble. This approach is appropriate when learning from massive-scale data that is too large to fit on a single machine. To get the best accuracy, IVoting should be used instead of bagging to generate the training subset for each decision tree in the random forest. Experiments with two large datasets (5GB and 50GB compressed) show that COMET compares favorably (in both accuracy and training time) to learning on a subsample of data using a serial algorithm. Finally, we propose a new Gaussian approach for lazy ensemble evaluation which dynamically decides how many ensemble members to evaluate per data point; this can reduce evaluation cost by 100X or more

Abstract-In this paper we present a novel approach for representing trajectories using sequenced linear dynamical systems. This method uses a closed-form least-squares procedure to fit a single Linear Dynamical System (LDS) to a simple trajectory. These LDS estimates form the elemental building blocks used to describe complicated trajectories through an automatic segmentation procedure that can represent complicated trajectories with high accuracy. Each estimated LDS induces a control law, mapping current state to desired state, that encodes the target trajectory in a generative manner. We provide a proof of stability of the control law and show how multiple trajectories can be incorporated to improve the generalization ability of the system.

To focus on the research issues surrounding collaborative behavior in multiple mobile-robotic systems, a great amount of low-level infrastructure is required. To facilitate our on-going research into multi-robot systems, we have developed RAVE, a software framework that provides a Real And Virtual Environment for running and managing multiple heterogeneous mobile-robot systems. This framework simplifies the implementation and development of collaborative robotic systems by providing the following capabilities: the 'ability to run systems off-line in simulation, user-interfaces for observing and commanding simulated and real robots, transparent transference of simulated robot programs to real robots, the ability to have simulated robots interact with real robots, and the ability to place virtual sensors on real robots to augment or experiment with their performance.

No abstract