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 20 of 41 for “"Energy Based Models"”.
-
Energy-based models for speech synthesis
… autoregressive (AR) and non-autoregressive (NAR) models that generate highly natural speech but can still produce inconsistencies. This thesis is motivated by the need to address these inconsistencies using generative models, particularly energy-based models (EBMs), and investigates how EBMs can …
-
Applications and Methods for Energy-based Models at Scale
Energy-Based Models (EBMs) are a class of generative models like Variational Autoencoders, Normalizing Flows, and Autoregressive Models. It is a commonly held belief that generative models like these can help to improve downstream discriminative machine learning applications. Generative models …
-
On the Learning of Energy-Based Models Using Noise Contrastive Estimation
Energy-Based Models (EBMs) are a family of unsupervised machine learning models that associate each point in the input space with an energy value in which low energy indicates a high likelihood. Specifically, an EBM can be viewed as an unnormalized probabilistic model, which upon normalization, …
-
Energy-Based Models for Medical Computer-Vision Applications in Low-Data Regimes.
… in both real-world and academic settings. Most models are trained in a supervised manner, aiming to replicate predefined information. While this approach achieves high accuracy, its application is limited, particularly in domains like life sciences, where data is abundant but often lacks …
-
Some statistical contributions to physics-informed neural networks and energy-based models
… on Physics-Informed Neural Networks (PINNs) and Energy-Based Models (EBMs). In the first part, the research examines Bayesian Physics-Informed Neural Networks (BPINNs), which integrate physical constraints directly into deep learning frameworks. This work investigates how BPINNs infer unknown …
-
Generative Models as Out-of-equilibrium particle systems: the case of Energy-Based Models
L'abstract è presente nell'allegato / the abstract is in the attachment
-
Evaluating Supplementation Strategies Using Precision Livestock Technologies and Energy Based Models in Developing Beef Heifers Grazing Dormant Rangeland
… of individual animal performance, while energy-based nutritional models support individualized supplementation. However, their application under extensive grazing conditions remains poorly understood. Therefore, the objectives of this thesis were to (1) evaluate heifer utilization of …
-
Compositional visual generation with energy-based modeling
Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2023-12-04 without embargo terms
-
Toward more scalable structured models
… biology and physical sciences, training such models is known to require significant amounts of data. One possible reason is that the structural properties of the data and problem are not modeled explicitly. Effectively exploiting the structure can help build more efficient and performing …
-
LEARNING DEEP REPRESENTATIONS: PLATFORM AND ALGORITHMS
… DEEP LEARNING PLATFORM CALLED PURINE IS DESIGNED BASED ON BI-GRAPH ABSTRACTION, PURINE ACCELERATE NEURAL NET TRAINING BY HARNESSING THE POWER OF MULTIPLE GPUS ON MULTIPLE MACHINES. IN THE ALGORITHMS PART, WE PROPOSED THE ?NETWORK IN NETWORK?(NIN) AND THE ?NEURAL DYNAMICAL SYSTEM?(NDS) ALGORITHMS. …
-
Are Particle-Based Methods the Future of Sampling in Joint Energy Models? A Deep Dive into SVGD and SGLD
… Variational Gradient Descent (SVGD) with Joint Energy Models (JEMs), comparing its performance to Stochastic Gradient Langevin Dynamics (SGLD). We incorporated a generative loss term with an entropy component to enhance diversity and a smoothing factor to mitigate numerical instability issues …
-
Automatic Speech Recognition without Transcribed Speech or Pronunciation Lexicons
… construction of phonemic pronunciation lexicons based on submodular selection of a small set of words for manual annotation, or words from other languages for which we have IPA pronunciations. We also explore techniques for training sequence-to-sequence models with very small amounts of data by …
-
Learning to Coordinate Efficiently through Multiagent Soft Q-Learning in the presence of Game-Theoretic Pathologies
… and modern strategies of training and evaluating energy-based models on learning to get a sense of whether the pitfall is due to sampling inefficiencies or underlying assumptions of the multiagent soft Q-learning extension (MASQL). We use the word sampler to refer to mechanisms that allow one to …
-
Active Perception for 3D Scene Representations in Robotics from an Information Theoretic Perspective
… taking an observation comes with a cost: time, energy, and risk. This dissertation explores the development of active robot perception systems utilizing radiance field representations, particularly 3D Gaussian Splatting, from an information-theoretic perspective. We first address the challenge …
-
Exploring Advanced Techniques for Better Generative Modeling
Deep generative models (DGMs) have wide applications in various aspects, e.g., Bayesian sampling, image or video generation and text generation. DGMs are typically learned to approximate complex and high-dimensional distributions, and then to generate samples that reside in the distributions. …
-
Cross-frequency coincidence detection in the processing of complex sounds
… previously believed to be difficult to explain. Models for psychophysical detection and discrimination were designed based on population responses of model coincidence cells. Quantitative predictions of masked detection suggest that the most sensitive model cells for detection are the cells whose …
-
Direct numerical simulations of bubbly flows
… considers direct numerical simulation (DNS) based studies of bubbly flows in turbulent channel and pipe flows. A variety of bubble sizes are simulated by means of Lagrangian tracking (LPT) and the volume of fluid (VOF) approaches to gain an in-depth knowledge of the physical phenomena …
-
Deep learning a transferable model for drug-receptor binding-energy
… particular, the construction of machine learning-based force fields for describing atomic interactions has been an attractive area of research because it may allow to close gaps between the efficiency of molecular dynamics simulations and accuracy of theoretical models. In this thesis, we explore …
-
Development of Multifunctional Shape Memory Polymer and Shape Memory Polymer Composites
… shape recovery abilities of an epoxy-based SMP. In doing so, epoxy based SMP is synthesized and thermo-mechanically characterized. Results show that the present SMP exhibits excellent shape recoveries under unconstraint conditions, for a range of fixing strains and temperatures. …
-
Dynamics of Drop Impact and Infiltration into Fur-Like Structures
… number as a predictor of capillary infiltration. Energy-based models are developed to predict penetration depth from above-array imaging, applicable to fibers of arbitrary convex cross section. Collectively, our findings unravel how the multi-functional design of natural and synthetic fibrous …
Page 1 of 3