Most Popular Books by Ian Andrew

This book''s hero is Darius Antonopoulos, an artist and apprentice printer who left London in 1605 to visit his ancestral home in Greece. What he found there was to send him on an incredible journey - a quest to discover the Seven Wonders of the World.

Gaussian processes are powerful Bayesian non-parametric models used for their closed-form posterior and marginal likelihoods. Yet, traditional methods for computing these quantities are not scalable to large data sets. And traditional kernels often do not express inductive biases that allow sample-efficient learning. Finding and exploiting structure in Gaussian processes builds a path towards tackling both problems. However, this has traditionally been accomplished through burdensome learning procedures, often by maximizing the marginal likelihood or variational objectives. In this thesis, we present methods for creating and exploiting structure in Gaussian processes directly. We first present randomly projected additive Gaussian processes, a class of Gaussian processes whose kernels operate additively over a set of random data projections. We prove that these kernels converge almost surely to a limiting kernel, which can be analytically derived in certain cases. We derive error bounds that characterize this convergence rate. We propose modifications to randomly projected additive Gaussian processes that improve their empirical convergence rate and regression performance. Next, we present algorithms for performing efficient online Gaussian process inference. Specifically, we present algorithms for computing the marginal likelihood, score function, and predictive distributions in constant time with respect to the number of observed data by using kernel interpolation approximations and algebraic manipulations using the Woodbury matrix identity.

This submission for the degree of PhD at the Elder Conservatorium of Music, University of Adelaide, is a portfolio of nineteen original compositions supported by an explanatory exegesis. These compositions are grouped into three major works: 1. Winds and Waters - a 53-minute song cycle for four singers and orchestra, comprising 13 individual songs, including solos, duets and a central quartet. 2. Nor the Storms That Pass - a 16-minute collection of five overlapping works for advanced a capella S.A.T.B vocal ensemble or choir. 3. Vespers - a 30-minute large-scale, non-liturgical setting of the Roman-Catholic Vespers mass for symphonic orchestra, chorus and vocal soloists. This project explores the compositional techniques involved in the development and realisation of original narrative-driven works by synthesising existing textual material from various unrelated sources, authors, historical eras and geographical locations into new and cohesive works with a perceptible storyline that was not necessarily present or implicit in any of the pre-existing works. This is achieved through an examination of the cognitive processes by which humans infer missing information from a partially-defined narrative, and the subsequent exploration and application of compositional techniques and treatments of synthesised texts which exploit this knowledge to most effectively guide a listener''s perceptions of textual cohesion. The works in the portfolio serve as a practical example of the application of the techniques being explored. The exegesis aims to provide technical analysis and insight into the applied creative process. The complete project serves as an educational resource for composers, writers, academics and professionals who have an interest in creating new works from existing materials or in understanding some of the compositional techniques which may be used to progress a narrative, and includes over 100 minutes of musical examples via the portfolio of works. The study also contributes a large body of new vocal and choral works to the repertoire, including thirteen new Australian art-songs, five new Australian choral works and a large-scale work for orchestra and chorus.