Determinantal Clustering Processes - A Nonparametric Bayesian Approach to Kernel Based Semi-Supervised Clustering

Semi-supervised clustering is the task of clustering data points into clusters where only a fraction of the points are labelled. The true number of clusters in the data is often unknown and most models require this parameter as an input. Dirichlet process mixture models are appealing as they can infer the number of clusters from the data. However, these models do not deal with high dimensional data well and can encounter difficulties in inference. We present a novel nonparameteric Bayesian kernel based method to cluster data points without the need to prespecify the number of clusters or to model complicated densities from which data points are assumed to be generated from. The key insight is to use determinants of submatrices of a kernel matrix as a measure of how close together a set of points are. We explore some theoretical properties of the model and derive a natural Gibbs based algorithm with MCMC hyperparameter learning. The model is implemented on a variety of synthetic and real world data sets.

Is performance measurement and management fit for the future?

Performance measurement and management (PMM) is a management and research paradox. On one hand, it provides management with many critical, useful, and needed functions. Yet, there is evidence that it can adversely affect performance. This paper attempts to resolve this paradox by focusing on the issue of "fit". That is, in today's dynamic and turbulent environment, changes in either the business environment or the business strategy can lead to the need for new or revised measures and metrics. Yet, if these measures and metrics are either not revised or incorrectly revised, then we can encounter situations where what the firm wants to achieve (as communicated by its strategy) and what the firm measures and rewards are not synchronised with each other (i.e., there is a lack of "fit"). This situation can adversely affect the ability of the firm to compete. The issue of fit is explored using a three phase Delphi approach. Initially intended to resolve this first paradox, the Delphi study identified another paradox - one in which the researchers found that in a dynamic environment, firms do revise their strategies, yet, often the PMM system is not changed. To resolve this second paradox, the paper proposes a new framework - one that shows that under certain conditions, the observed metrics "lag" is not only explainable but also desirable. The findings suggest a need to recast the accepted relationship between strategy and PMM system and the output included the Performance Alignment Matrix that had utility for managers. © 2013 .