Using Bayesian neural networks to classify segmented images

We present results that compare the performance of neural networks trained with two Bayesian methods, (i) the Evidence Framework of MacKay (1992) and (ii) a Markov Chain Monte Carlo method due to Neal (1996) on a task of classifying segmented outdoor images. We also investigate the use of the Automatic Relevance Determination method for input feature selection.

Social psychological aspects of rape: a study of recorded rape and images of rape victimisation

This thesis is organised into four parts. In Part 1 relevant literature is reviewed and presented in three chapters. Chapter 1 examines legal and cultural factors in identifying the. boundaries of rape. Chapter 2 discusses idiographic features· and causal characteristics of rape suspects and victims. Chapter 3 reviews the evidence relating to attitudes toward rape,. attribution of responsibility to victims and the routine management of rape cases by the police. Part II comprises an experimental investigation of observer perception of the victims of violent crime. The experiment, examined the processes by which impressions were attributed to victims of personal crime. The results suggested that discrepancies from observers' stereotypes were an important factor in their polarisation of victim ratings. The relevance of. examining . both the structure and process of' impression, formation was highlighted. Part III describes an extensive field study in which the West. Midlands police files on rape for an eight year period (1071-1978) were analysed. The study revealed a large number of interesting findings related to a wide range of relevant features of the crime. Further, the impact .of common misconceptions and "myths" of rape were investigated across the legal and judicial processing of rape cases. The evidence suggests that these "myths" lead·to differential biasing effects at different stages in the process. In the final part of this thesis,. salient issues raised by the experiment and field study .are discussed·within the framework outlined in Part 1. Potential implications for future developments and research: are presented.

Sparse representation of astronomical images

Sparse representation of astronomical images is discussed. It is shown that a significant gain in sparsity is achieved when particular mixed dictionaries are used for approximating these types of images with greedy selection strategies. Experiments are conducted to confirm (i) the effectiveness at producing sparse representations and (ii) competitiveness, with respect to the time required to process large images. The latter is a consequence of the suitability of the proposed dictionaries for approximating images in partitions of small blocks. This feature makes it possible to apply the effective greedy selection technique called orthogonal matching pursuit, up to some block size. For blocks exceeding that size, a refinement of the original matching pursuit approach is considered. The resulting method is termed "self-projected matching pursuit," because it is shown to be effective for implementing, via matching pursuit itself, the optional backprojection intermediate steps in that approach. © 2013 Optical Society of America.