We present a new approach to detecting defects in random textures which requires only very few defect free samples for unsupervised training. Each product image is divided into overlapping patches of various sizes. Then, density mixture models are applied to reduce groupings of patches to a number of textural exemplars, referred to here as texems, characterising the means and covariances of who...

A novelty detection approach to condition monitoring of aerospace gas-turbine engines is presented, providing a consistent framework for onand off-line analysis, each with differing typical implementation constraints. On-line techniques are introduced for observing abnormality in engine behaviour during aircraft flights, and are shown to provide early warning of engine events in real-time. Off-...

Application of novelty detection to a new class of jet engine is considered within this paper, providing a worked example of the steps necessary for constructing a model of normality. Abnormal jet engine vibration signatures are automatically identified with respect to a training set of normal examples. Pre-processing steps suitable for this area of application are investigated. An intuitive me...

Timeline summarization aims at generating concise summaries and giving readers a faster and better access to understand the evolution of news. It is a new challenge which combines salience ranking problem with novelty detection. Previous researches in this field seldom explore the evolutionary pattern of topics such as birth, splitting, merging, developing and death. In this paper, we develop a...

Emergence has become the focus of much attention in the design computing research in recent years as one of only a small number of design processes that can expand a design space and hence facilitate creative designing as defined by Gero (1990). This poster presentation examines the requirements for a computational model of design emergence and suggests that many existing computational systems ...

Rumour detection is hard because the most accurate systems operate retrospectively, only recognising rumours once they have collected repeated signals. By then the rumours might have already spread and caused harm. We introduce a new category of features based on novelty, tailored to detect rumours early on. To compensate for the absence of repeated signals, we make use of news wire as an addit...

Novelty detection is the ident ification of new or unknown data or signal that a machine learning system is not aware of during training. In this paper we focus on neural network based approaches for novelty detection. Statistical approaches are covered in part-I paper.

This paper presents the preliminary results of an observational study into the use of novelty detection techniques for detecting physiological deterioration in vital-sign data acquired from Emergency Department (ED) patients. Such patients are typically in an acute condition with a significant chance of deteriorating during their stay in hospital. Existing methods for monitoring ED patients inv...