Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 15 of 15 for “"model perturbation"”.
-
Generalization in Machine Learning Through Information-Theoretic Lens
… we apply information-theoretic analysis for models trained using stochastic gradient descent (SGD). We do so by invoking an auxiliary weight process and by approximating SGD using stochastic differential equations (SDE). Our analysis reveals intriguing phenomena such as epoch-wise double …
-
Functional fault modeling and test vector development for VLSI systems
… at the same pace as the technology. The fault-models proposed for LSI systems are no longer capable of testing VLSI devices efficiently. Thus the stuck-at and short/open fault models are outdated. Despite this fact, these old models are used in the industry with some modifications. Also, these …
-
Geometric Aspects of Type IIA Supersymmetric Backgrounds and Heterotic Anomalies
… the algebra of holonomy symmetries of sigma models propagating on supersymmetric heterotic backgrounds with a non-compact holonomy group is determined. One demonstrates that these close as a W-algebra that is specified by a Lie algebra structure on the space of covariantly constant forms that …
-
Application of wavelet transforms to seismic data processing and inversion
… underdetermined and ill-posed. An initial model and some prior information describing the model is needed to find the model perturbation that minimizes an objective function. Traditionally, the perturbation is assumed to be an independent Gaussian or a correlated but stationary process, an …
-
Towards Efficient Self-Supervised Learning on Graphs
… for graphs primarily concentrate on enhancing model performance through advanced data augmentation strategies and contrastive loss functions. Despite the significant progress made by existing studies, they encounter severe efficiency challenges when dealing with large-scale graphs and …
-
A study of interpretability mechanisms for deep networks
… are traditionally considered to be “black-box” models where it is generally difficult to interpret a certain decision made by such models given a test instance. However, as deep learning is increasingly becoming the tool of choice in making many safety-critical and time-critical decisions such …
-
Ensemble control of robotic systems
… the plate-ball manipulator---in the presence of model perturbation that scales all inputs by an unknown but bounded constant. The basic idea is to maintain the set of all possible configurations and to select inputs that reduce the size of this set and drive it toward some goal configuration. The …
-
Feedback control of polymer flooding process considering geologic uncertainty
… dynamically controlled using updated reservoir models and a closed-loop production optimization scheme is implemented. However, the formulation of an optimal production strategy is based on uncertain production forecasts resulting from uncertainty in spatial representation of reservoir …
-
THE POPULATION BIOLOGY AND ECOSYSTEM EFFECTS OF THE SEA NETTLE, CHRYSAORA CHESAPEAKEI
… methods was analyzed. Medusa abundance was modeled by change points and was highly correlated between the sampling methods. The remaining signal was random, and indices indicated that medusae were aggregated. This study suggests more monitoring from visual shore-based surveys is an …
-
A Perturbation Approach to Differential Privacy for Deep Learning based Speech Processing
… embarks on the journey of developing a perturbation-based machine learning framework, ensuring that DNNs maintain speech model performance whilst satisfying DP constraints. We devise mechanisms for vector-space distortion, achieved by introducing bounded noise to speech training data, …
-
Inverse optimal control for differentially flat systems with application to lower-limb prosthetic devices
… learning controllers for stair ascent, we need a model of locomotion that is capable of generating exemplar trajectories for any desired stair height. Motivated by this challenge, we focus on a class of learning from demonstration methods called inverse optimal control. Inverse optimal control is …
-
Efficient Computation of Boundary-Layer Instabilities in Highly Three-Dimensional Flows
… effects requires robust tools capable of modeling the underlying physical mechanisms that lead to turbulence. Transition is notoriously sensitive to the details of the underlying flow; therefore, any reduced-order models must balance correct physical assumptions with rising computational …
-
Voter Models On Graphs
<p>The voter model which describes the flow of information through interactions between neighbors has been widely studied in the field of probability. In this paper we study two variations of the voter model, one is called the Latent Voter Model and the other is called the Zealot Voter Model. Both …
-
Cost effective strategies for problems in computational geophysics : seismic modeling and imaging
The first part of my thesis focuses on seismic modeling in fractured media. Several recent developments in finite elements such as usage of high degree polynomials to approximate the wavefield, diagonalization of mass matrices to be inverted through mass lumping techniques and usage of high order …
-
Dynamic Limits of Balance Control during Daily Functional Activities Associated with Falling
… this dissertation was to develop biomechanical models that identify dynamic limits of balance control in daily functional activities associated with falling, including sit-to-stand (STS) movement, standing (stance perturbation), and walking. Poor performance of STS movement has been identified …