The simplicity principle posits that we interpret sense data as the simplest consistent distal cause, or that our high level perceptual representations of stimuli are optimized for simplicity. The traditional paradigm used to test this principle is coding theory, where alternate representations of stimuli are constructed, simplicity is measured as shortness of representation length, and behavioural experiments attempt to show that the shortest representations correspond best to perception. In this study we apply coding theory to marker-based human walking motion. We compare two representation schemes. The first is based on marker coordinates in a body-centred Cartesian coordinate system. The second is based on a model of 15 rigid body segments with Euler angles and a Cartesian translation for each. Both of our schemes are principal component (PC)-based implementations of a norm-based multidimensional object space – a type of model for high level perceptual schemes that has received attention in the literature over the past two decades. Representation length is quantified as number of retained PC’s, with error increasing with discarded PC’s. We generalize simplicity to efficiency measured as error across all possible lengths, where more efficient schemes admit less error across lengths. We find that the Cartesian coordinates-based scheme is more efficient than the Euler angles and translations-based scheme across a database of 100 walkers. In order to link this finding to perception we turn to the caricature effect that subjects can identify caricatures of familiar stimuli more accurately than veridicals. Our design was to compare walker caricatures generated in our two schemes in the hope of finding that one gives caricatures that benefit identification more than the other, from which we would conclude the former to be a better model of the true perceptual scheme. However, we find that analogous caricatures between the two schemes are only distinguishable at caricature levels so extreme that identification performance breaks down, so our design became infeasible and no conclusion for a simplicity principle in walker perception is reached. We also measure a curve of increasing then decreasing identification performance with caricature level and an optimal level at approximately double the distinctiveness of a typical walker. / Thesis (Master, Neuroscience Studies) -- Queen's University, 2010-10-29 19:16:39.943

Two types of artificial neural networks, Multilayer Perceptron (MLP) and Self-organizing Feature Map (SOM), were employed to detect mastitis for robotic milking stations using the preprocessed data relating to the electrical conductivity and milk yield. The SOM was developed to classify the health status into three categories: healthy, moderately ill and severely ill. The clustering results were successfully evaluated and validated by using statistical techniques such as K-means clustering, ANOVA and Least Significant Difference. The result shows that the SOM could be used in the robotic milking stations as a detection model for mastitis. For developing MLP models, a new mastitis definition based on higher EC and lower quarter yield was created and Principle Components Analysis technique was adopted for addressing the problem of multi-colinearity existed in the data. Four MLPs with four combined datasets were developed and the results manifested that the PCA-based MLP model is superior to other non-PCA-based models in many respects such as less complexity, higher predictive accuracy. The overall correct classification rate (CCR), sensitivity and specificity of the model was 90.74 %, 86.90 and 91.36, respectively. We conclude that the PCA-based model developed here can improve the accuracy of prediction of mastitis by robotic milking stations.