IGImène R. Goumiri

Imène R. Goumiri

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Imène R. Goumiri has a total of 20 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Eucalyptus – An Analysis Suite for Fault Trees with Uncertainty Quantification

    by Imene Goumiri, Luc Peterson, Adam G Taylor, Jayson “Luc” Peterson

    This paper introduces Eucalyptus, a new software suite for fault tree analysis and uncertainty quantification to enhance reliability engineering.

  • Open Access

    MuyGPs: Scalable Gaussian Process Hyperparameter Estimation Using Local Cross-Validation

    by Amanda Muyskens, Benjamin Priest, Imène Goumiri, Michael Schneider

    Gaussian processes (GPs) are non-linear probabilistic models popular in many applications. However, na\"ive GP realizations require quadratic memory to store the covariance matrix and cubic computation to perform inference or evaluate the likelihood function. These bottlenecks have driven much investment in the development of approximate GP alternatives that scale to the large data sizes common in modern data-driven applications. We present in this manuscript MuyGPs, a novel efficient GP hyperparameter estimation method. MuyGPs builds upon prior methods that take advantage of the nearest neighbors structure of the data, and uses leave-one-out cross-validation to optimize covariance (kernel) hyperparameters without realizing a possibly expensive likelihood. We describe our model and methods in detail, and compare our implementations against the state-of-the-art competitors in a benchmark spatial statistics problem. We show that our method outperforms all known competitors both in terms of time-to-solution and the root mean squared error of the predictions.


  • Simultaneous feedback control of toroidal magnetic field and plasma current on MST using advanced programmable power supplies

    by I R Goumiri, K J McCollam, A A Squitieri, D J Holly, J S Sarff, S P Leblanc

    Programmable control of the inductive electric field enables advanced operations of reversed-field pinch (RFP) plasmas in the Madison Symmetric Torus (MST) device and further develops the technical basis for ohmically heated fusion RFP plasmas. MST’s poloidal and toroidal magnetic fields ( B p and B t ) can be sourced by programmable power supplies (PPSs) based on integrated-gate bipolar transistors (IGBT). In order to provide real-time simultaneous control of both B p and B t circuits, a time-independent integrated model is developed. The actuators considered for the control are the B p and B t primary currents produced by the PPSs. The control system goal will be tracking two particular demand quantities that can be measured at the plasma surface ( r = a ): the plasma current, I p ∼ B p ( a ), and the RFP reversal parameter, F ∼ B t ( a )/Φ, where Φ is the toroidal flux in the plasma. The edge safety factor, q ( a ) ∝ B t ( a ), tends to track F but not identically. To understand the responses of I p and F to the actuators and to enable systematic design of control algorithms, dedicated experiments are run in which the actuators are modulated, and a linearized dynamic data-driven model is generated using a system identification method. We perform a series of initial real-time experiments to test the designed feedback controllers and validate the derived model predictions. The feedback controllers show systematic improvements over simpler feedforward controllers.


  • Open Access

    Simultaneous feedback control of toroidal magnetic field and plasma current on MST using advanced programmable power supplies

    by I. R. Goumiri, K. J. McCollam, A. A. Squitieri, D. J. Holly, J. S. Sarff, S. P. Leblanc

    Programmable control of the inductive electric field enables advanced operations of reversed-field pinch (RFP) plasmas in the Madison Symmetric Torus (MST) device and further develops the technical basis for ohmically heated fusion RFP plasmas. MST's poloidal and toroidal magnetic fields ($B_\text{p}$ and $B_\text{t}$) can be sourced by programmable power supplies (PPSs) based on integrated-gate bipolar transistors (IGBT). In order to provide real-time simultaneous control of both $B_\text{p}$ and $B_\text{t}$ circuits, a time-independent integrated model is developed. The actuators considered for the control are the $B_\text{p}$ and $B_\text{t}$ primary currents produced by the PPSs. The control system goal will be tracking two particular demand quantities that can be measured at the plasma surface ($r=a$): the plasma current, $I_\text{p} \sim B_\text{p}(a)$, and the RFP reversal parameter, $F\sim B_\text{t}(a)/\Phi$, where $\Phi$ is the toroidal flux in the plasma. The edge safety factor, $q(a)\propto B_t(a)$, tends to track $F$ but not identically. To understand the responses of $I_\text{p}$ and $F$ to the actuators and to enable systematic design of control algorithms, dedicated experiments are run in which the actuators are modulated, and a linearized dynamic data-driven model is generated using a system identification method. We perform a series of initial real-time experiments to test the designed feedback controllers and validate the derived model predictions. The feedback controllers show systematic improvements over simpler feedforward controllers.

  • Open Access

    Reinforcement Learning via Gaussian Processes with Neural Network Dual Kernels

    by Imène R. Goumiri, Benjamin W. Priest, Michael D. Schneider

    While deep neural networks (DNNs) and Gaussian Processes (GPs) are both popularly utilized to solve problems in reinforcement learning, both approaches feature undesirable drawbacks for challenging problems. DNNs learn complex nonlinear embeddings, but do not naturally quantify uncertainty and are often data-inefficient to train. GPs infer posterior distributions over functions, but popular kernels exhibit limited expressivity on complex and high-dimensional data. Fortunately, recently discovered conjugate and neural tangent kernel functions encode the behavior of overparameterized neural networks in the kernel domain. We demonstrate that these kernels can be efficiently applied to regression and reinforcement learning problems by analyzing a baseline case study. We apply GPs with neural network dual kernels to solve reinforcement learning tasks for the first time. We demonstrate, using the well-understood mountain-car problem, that GPs empowered with dual kernels perform at least as well as those using the conventional radial basis function kernel. We conjecture that by inheriting the probabilistic rigor of GPs and the powerful embedding properties of DNNs, GPs using NN dual kernels will empower future reinforcement learning models on difficult domains.

  • The effect of capillary pressure on the saturation equation of two‐phase flow in porous media

    by Imène R. Goumiri, Jean H. Prévost, Matthias Preisig

    SUMMARY: A complete and accurate simulation of two‐phase flow in porous media requires knowledge of all the controlling physics (and values of physical parameters) that play a relevant role and an understanding of the effects of each one on the solution. Of particular concern here is the effect of capillary pressure and the length scale over which it is relevant. The goal of this paper is to provide guidance onto when to include the effects of capillary pressure in the model, and onto what are the resulting length scale restrictions if those effects are to be included. Copyright © 2011 John Wiley & Sons, Ltd.