Frank and Stein : from research scientist to creative writer (a screenplay and exegesis)

The screenplay, “Perfect Blood” (Frank and Stein), is the first two-hour episode of a two-part television miniseries Frank and Stein. This creative work is a science fiction story that speculates on the future of Western nations in a world where petroleum is scarce. A major theme that has been explored in the miniseries is the tension between the advantages and dangers of scientific progress without regard to human consequences. “Perfect Blood” (Frank and Stein) was written as part of my personal creative journey, which has been the transformation from research scientist to creative writer. In the exegetical component of this thesis, I propose that a key challenge for any scientist writing science fiction is the shift from conducting empirical research in a laboratory-based situation to engaging in creative practice research. During my personal creative journey, I found that a predominant difficulty in conducting research within a creative practice-led paradigm was unleashing my creativity and personal viewpoint, practices that are frowned upon in scientific research. The aim of the exegesis is to demonstrate that the transformative process from science to art is not neat and well-structured. My personal creative journey was fraught with many ‘wrong’ turns. However, after reflecting on the experience, I realise that every varied piece of research that I undertook allowed me to progress to the next stage, the next draft of Frank and Stein. And via the disorder of the creative process, a screenplay finally emerged that was both structured and creative, which are equally essential elements in screenwriting.

Soundage : a practice-led approach to Gertrude Stein, sound, and generative language

In this practice-led research project I work to show how a re-reading and a particular form of listening to the sound-riddled nature of Gertrude Stein's work, Two: Gertrude Stein and her Brother, presents us with a contemporary theory of sound in language. This theory, though in its infancy, is a particular enjambment of sounded language that presents itself as an event, engaged with meaning, with its own inherent voice. It displays a propensity through engagement with the 'other' to erupt into love. In this thesis these qualities are reverberated further through the work of Seth Kim-Cohen's notion of the non-cochlear, Simon Jarvis's notion of musical thinking, Jean-Jacques Lecercle's notion of délire or nonsense, Luce Irigaray's notion of jouissant love and the Bracha Ettinger's notion of the generative matrixial border space. This reading then is simultaneously paired with my own work of scoring and creating a digital opera from Stein's work, thereby testing and performing Stein's theory. In this I show how a re-reading and relistening to Stein's work can be significant to feminist ethical language frames, contemporary philosophy, sonic art theory and digital language frames. Further significance of this study is that when the reverberation of Stein's engagements with language through sound can be listened to, a pattern emerges, one that encouragingly problematizes subjectivity and interweaves genres/methods and means, creating a new frame for sound in language, one with its own voice that I call soundage.

Multi-shot person re-identification via relational stein divergence

Person re-identification is particularly challenging due to significant appearance changes across separate camera views. In order to re-identify people, a representative human signature should effectively handle differences in illumination, pose and camera parameters. While general appearance-based methods are modelled in Euclidean spaces, it has been argued that some applications in image and video analysis are better modelled via non-Euclidean manifold geometry. To this end, recent approaches represent images as covariance matrices, and interpret such matrices as points on Riemannian manifolds. As direct classification on such manifolds can be difficult, in this paper we propose to represent each manifold point as a vector of similarities to class representers, via a recently introduced form of Bregman matrix divergence known as the Stein divergence. This is followed by using a discriminative mapping of similarity vectors for final classification. The use of similarity vectors is in contrast to the traditional approach of embedding manifolds into tangent spaces, which can suffer from representing the manifold structure inaccurately. Comparative evaluations on benchmark ETHZ and iLIDS datasets for the person re-identification task show that the proposed approach obtains better performance than recent techniques such as Histogram Plus Epitome, Partial Least Squares, and Symmetry-Driven Accumulation of Local Features.