Probabilistic multidimensional scaling using a city-block metric (Q1599153)

Exact and approximate, computationally efficient Probability Density Functions (PDFs) for city-block distances and distance ratios are considered. The main assumption of the presented probabilistic model is that stimuli can be represented by random vectors having multivariate normal distributions. As the stimulus variance magnitudes increase, the differences between city-block and Euclidean PDFs diminish. The problem of distinguishing a city-block or Euclidean metric under different levels of stimulus variability by models based upon city-block PDFs is briefly discussed.