Attention vs. emotion: Affective learning and the startle reflex

The eye-blink startle reflex can be modulated by attentional and emotional processes. The reflex is facilitated during stimuli that engage attention. A linear pattern of emotional modulation has also been consistently demonstrated: the reflex is facilitated during unpleasant stimuli and attenuated during pleasant stimuli. However, during anticipation of pleasant or unpleasant stimuli it is unclear whether emotion or attention drives startle reflex modulation. This study used a differential learning procedure to investigate whether startle modulation during anticipation of a salient stimulus reflected emotional or attentional processes. In acquisition, a CS+ was paired with a pleasant or unpleasant US and a CS- was presented alone. In extinction, blink startle magnitude was measured during CS+ and CS-. Post-acquisition valence ratings and affective priming showed that CS+ had acquired the same affective value as the pleasant or unpleasant US with which it was paired. No differences in modulation of blink startle reflexes during pleasant CS+ and unpleasant CS+ were found throughout extinction. Blink startle facilitation occurred during CS+ but not CS- across the first third of extinction. Thus, attentional rather than emotional processes appeared to facilitate blink startle during anticipation of salient stimuli.

Machine learning for matching astronomy catalogues

An emerging issue in the field of astronomy is the integration, management and utilization of databases from around the world to facilitate scientific discovery. In this paper, we investigate application of the machine learning techniques of support vector machines and neural networks to the problem of amalgamating catalogues of galaxies as objects from two disparate data sources: radio and optical. Formulating this as a classification problem presents several challenges, including dealing with a highly unbalanced data set. Unlike the conventional approach to the problem (which is based on a likelihood ratio) machine learning does not require density estimation and is shown here to provide a significant improvement in performance. We also report some experiments that explore the importance of the radio and optical data features for the matching problem.

Real-Time Gene-Expression Extraction with Fast Classification (FC) Networks

Fast Classification (FC) networks were inspired by a biologically plausible mechanism for short term memory where learning occurs instantaneously. Both weights and the topology for an FC network are mapped directly from the training samples by using a prescriptive training scheme. Only two presentations of the training data are required to train an FC network. Compared with iterative learning algorithms such as Back-propagation (which may require many hundreds of presentations of the training data), the training of FC networks is extremely fast and learning convergence is always guaranteed. Thus FC networks may be suitable for applications where real-time classification is needed. In this paper, the FC networks are applied for the real-time extraction of gene expressions for Chlamydia microarray data. Both the classification performance and learning time of the FC networks are compared with the Multi-Layer Proceptron (MLP) networks and support-vector-machines (SVM) in the same classification task. The FC networks are shown to have extremely fast learning time and comparable classification accuracy.

Learning to play Pac-Man: An evolutionary, rule-based approach

Pac-Man is a well-known, real-time computer game that provides an interesting platform for research. We describe an initial approach to developing an artificial agent that replaces the human to play a simplified version of Pac-Man. The agent is specified as a simple finite state machine and ruleset. with parameters that control the probability of movement by the agent given the constraints of the maze at some instant of time. In contrast to previous approaches, the agent represents a dynamic strategy for playing Pac-Man, rather than a pre-programmed maze-solving method. The agent adaptively "learns" through the application of population-based incremental learning (PBIL) to adjust the agents' parameters. Experimental results are presented that give insight into some of the complexities of the game, as well as highlighting the limitations and difficulties of the representation of the agent.

Playing in continuous spaces: Some analysis and extension of population-based incremental learning

As an alternative to traditional evolutionary algorithms (EAs), population-based incremental learning (PBIL) maintains a probabilistic model of the best individual(s). Originally, PBIL was applied in binary search spaces. Recently, some work has been done to extend it to continuous spaces. In this paper, we review two such extensions of PBIL. An improved version of the PBIL based on Gaussian model is proposed that combines two main features: a new updating rule that takes into account all the individuals and their fitness values and a self-adaptive learning rate parameter. Furthermore, a new continuous PBIL employing a histogram probabilistic model is proposed. Some experiments results are presented that highlight the features of the new algorithms.

Representation and learning of visual information for pose recognition

Recovering position from sensor information is an important problem in mobile robotics, known as localisation. Localisation requires a map or some other description of the environment to provide the robot with a context to interpret sensor data. The mobile robot system under discussion is using an artificial neural representation of position. Building a geometrical map of the environment with a single camera and artificial neural networks is difficult. Instead it would be simpler to learn position as a function of the visual input. Usually when learning images, an intermediate representation is employed. An appropriate starting point for biologically plausible image representation is the complex cells of the visual cortex, which have invariance properties that appear useful for localisation. The effectiveness for localisation of two different complex cell models are evaluated. Finally the ability of a simple neural network with single shot learning to recognise these representations and localise a robot is examined.

Using ancillary statistics in on-line learning algorithms

Neural networks are usually curved statistical models. They do not have finite dimensional sufficient statistics, so on-line learning on the model itself inevitably loses information. In this paper we propose a new scheme for training curved models, inspired by the ideas of ancillary statistics and adaptive critics. At each point estimate an auxiliary flat model (exponential family) is built to locally accommodate both the usual statistic (tangent to the model) and an ancillary statistic (normal to the model). The auxiliary model plays a role in determining credit assignment analogous to that played by an adaptive critic in solving temporal problems. The method is illustrated with the Cauchy model and the algorithm is proved to be asymptotically efficient.

Incorporating curvature information into on-line learning

We analyse the dynamics of a number of second order on-line learning algorithms training multi-layer neural networks, using the methods of statistical mechanics. We first consider on-line Newton's method, which is known to provide optimal asymptotic performance. We determine the asymptotic generalization error decay for a soft committee machine, which is shown to compare favourably with the result for standard gradient descent. Matrix momentum provides a practical approximation to this method by allowing an efficient inversion of the Hessian. We consider an idealized matrix momentum algorithm which requires access to the Hessian and find close correspondence with the dynamics of on-line Newton's method. In practice, the Hessian will not be known on-line and we therefore consider matrix momentum using a single example approximation to the Hessian. In this case good asymptotic performance may still be achieved, but the algorithm is now sensitive to parameter choice because of noise in the Hessian estimate. On-line Newton's method is not appropriate during the transient learning phase, since a suboptimal unstable fixed point of the gradient descent dynamics becomes stable for this algorithm. A principled alternative is to use Amari's natural gradient learning algorithm and we show how this method provides a significant reduction in learning time when compared to gradient descent, while retaining the asymptotic performance of on-line Newton's method.

Transients and asymptotics of natural gradient learning

We analyse natural gradient learning in a two-layer feed-forward neural network using a statistical mechanics framework which is appropriate for large input dimension. We find significant improvement over standard gradient descent in both the transient and asymptotic phases of learning.