Aqsa Shabbir

ELM Regime Classification by Conformal Prediction on an Information Manifold

Characterization and control of plasma instabilities known as edge-localized modes (ELMs) is crucial for the operation of fusion reactors. Recently, machine learning methods have demonstrated good potential in making useful inferences from stochastic fusion data sets. However, traditional classification methods do not offer an inherent estimate of the goodness of their prediction. In this paper, a distance-based conformal predictor classifier integrated with a geometric-probabilistic framework is presented. The first benefit of the approach lies in its comprehensive treatment of highly stochastic fusion data sets, by modeling the measurements with probability distributions in a metric space. This enables calculation of a natural distance measure between probability distributions: the Rao geodesic distance. Second, the predictions are accompanied by estimates of their accuracy and reliability. The method is applied to the classification of regimes characterized by different types of ELMs based on the measurements of global parameters and their error bars. This yields promising success rates and outperforms state-of-the-art automatic techniques for recognizing ELM signatures. The estimates of goodness of the predictions increase the confidence of classification by ELM experts, while allowing more reliable decisions regarding plasma control and at the same time increasing the robustness of the control system.

Multivariate texture discrimination based on geodesics to class centroids on a generalized Gaussian Manifold

A texture discrimination scheme is proposed wherein probability distributions are deployed on a probabilistic manifold for modeling the wavelet statistics of images. We consider the Rao geodesic distance (GD) to the class centroid for texture discrimination in various classification experiments. We compare the performance of GD to class centroid with the Euclidean distance in a similar context, both in terms of accuracy and computational complexity. Also, we compare our proposed classification scheme with the k-nearest neighbor algorithm. Univariate and multivariate Gaussian and Laplace distributions, as well as generalized Gaussian distributions with variable shape parameter are each evaluated as a statistical model for the wavelet coefficients. The GD to the centroid outperforms the Euclidean distance and yields superior discrimination compared to the k-nearest neighbor approach.

Multivariate texture discrimination using a principal geodesic classifier

A new texture discrimination method is presented for classification and retrieval of colored textures represented in the wavelet domain. The interband correlation structure is modeled by multivariate probability models which constitute a Riemannian manifold. The presented method considers the shape of the class on the manifold by determining the principal geodesic of each class. The method, which we call principal geodesic classification, then determines the shortest distance from a test texture to the principal geodesic of each class. We use the Rao geodesic distance (GD) for calculating distances on the manifold. We compare the performance of the proposed method with distance-to-centroid and k-nearest neighbor classifiers and of the GD with the Euclidean distance. The principal geodesic classifier coupled with the GD yields better results, indicating the usefulness of effectively and concisely quantifying the variability of the classes in the probabilistic feature space.

EFD-C(13)03/17 Visualization of Tokamak Operational Spaces Through the Projection of Data Probability Distributions

Information visualization is becoming an increasingly important tool for making inferences from large and complex data sets describing tokamak operational spaces. Landmark MDS, a computationally efficient information visualization tool, well suited to the properties of fusion data, along with a comprehensive probabilistic data representation framework, is shown to provide a structured visual map of plasma confinement regimes, plasma disruption regions and plasma trajectories. This is aimed at contributing to the understanding of underlying physics of various plasma phenomena, while providing an intuitive tool for plasma monitoring.