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 22 for “"Evidence lower bound"”.

  1. Stochastic Variational Autoencoder

    … the VAE framework, we aim to achieve a tighter evidence lower bound (ELBO) on the log-likelihood of the data. Our approach is a preliminary investigation to enhance the latent space structure and improve generative performance by incorporating richer uncertainty modeling.

    rit Repository record for Stochastic Variational Autoencoder (opens in a new tab)

  2. Essays on variational Bayes in Econometrics

    … VAR). The algorithm is derived based on the evidence lower bound, which is demonstrated to be tight, ensuring efficient convergence. The optimization is carried through the Coordinate descent optimization. To validate the proposed method, its accuracy and computational costs are compared with …

    strathclyde Repository record for Essays on variational Bayes in Econometrics (opens in a new tab)

  3. Understanding Variational Autoencoders and Disentanglement Metrics

    … the rich mathematical basis of VAEs, the evidence lower bound (ELBO) and how regularization of original VAEs brings the trade-off between reconstruction fidelity and quality of disentanglement within the learnt representations. We validate the theory by conducting experiments on Frey's …

    washington Repository record for Understanding Variational Autoencoders and Disentanglement Metrics (opens in a new tab)

  4. Variational learning for latent Gaussian model of discrete data

    … We consider a variational approach based on evidence lower bound optimization. We solve the following two main problems of the variational approach: the computational inefficiency associated with the maximization of the lower bound and the intractability of the lower bound. For the first …

    ubc Repository record for Variational learning for latent Gaussian model of discrete data (opens in a new tab) Download PDF for Variational learning for latent Gaussian model of discrete data (opens in a new tab)

  5. Symmetric Variational Inference with High Mutual Information

    … inference stands in contrast to the use of an evidence-lower-bound (ELBO) in VAEs, and the use of adversarial learning that is common with other models formulated in terms of a symmetric divergence. In this thesis we systematically probe different terms in the variational bound, providing …

    toronto-retro Repository record for Symmetric Variational Inference with High Mutual Information (opens in a new tab)

  6. Methodological advancements for synthetic data generation

    … options and a distributionally weighted evidence lower bound (wTAB-ELBO). In the second project, we present Distributionally Aware Tabular VAE (DA-TVAE), an extension of the AutoTab framework that introduces variable-type specific distributional choices, including extensions to count, …

    bu Repository record for Methodological advancements for synthetic data generation (opens in a new tab)

  7. Gradient estimation for discrete variables via dependent Monte Carlo samples

    … latent variables, trained using either the evidence lower bound or the multi sample bound. We also compare our approach on a structured prediction task for training stochastic categorical networks. Lastly, we evaluate ARMS for different Plackett-Luce based objectives, which include …

    texas Repository record for Gradient estimation for discrete variables via dependent Monte Carlo samples (opens in a new tab)

  8. Generative Bayesian co-clustering for Parkinson’s disease subtyping using noisy multimodal imaging and clinical data

    … on rows and columns. A doubly re‑parameterized evidence lower bound (ELBO) sharpens gradient signal‑to‑noise separation, while a regularized Compositional ELBO simultaneously reconstructs input data and learns an explicit noise model, allowing robust training in the presence of missing or …

    texas Repository record for Generative Bayesian co-clustering for Parkinson’s disease subtyping using noisy multimodal imaging and clinical data (opens in a new tab)

  9. Asymptotically exact variational inference via measure-preserving dynamical systems

    … that errors in sampling, density evaluation, and evidence lower bound estimation grow far more slowly than trajectory errors themselves, which often grow exponentially with flow length.

    ubc Repository record for Asymptotically exact variational inference via measure-preserving dynamical systems (opens in a new tab) Download PDF for Asymptotically exact variational inference via measure-preserving dynamical systems (opens in a new tab)

  10. Advancing variational inference via thermodynamic integration

    … via optimization. Central to VI is the Evidence Lower Bound (ELBO), a variational objective function which lower bounds the log marginal likelihood, and can be used to jointly perform maximum likelihood parameter estimation and approximate posterior inference using stochastic gradient …

    ubc Repository record for Advancing variational inference via thermodynamic integration (opens in a new tab) Download PDF for Advancing variational inference via thermodynamic integration (opens in a new tab)

  11. Approximate Inference in Variational Autoencoders

    … autoencoders is that they maximize a tighter lower bound on the marginal likelihood than the standard evidence lower bound. The first contribution of this thesis is to provide an alternative interpretation: that it optimizes the standard variational lower bound, but using a stochastic …

    toronto-retro Repository record for Approximate Inference in Variational Autoencoders (opens in a new tab)

  12. Generative models: theory and applications

    … accuracy of diffusion models. We derive error bounds for flow matching - a generalization of diffusion models - and improve upon state-of-the-art bounds for diffusion models, using techniques inspired by stochastic localization.</p> <p>Importance weighted autoencoders (IWAEs) work by learning a …

    oxford Repository record for Generative models: theory and applications (opens in a new tab)

  13. Geometric Methods in MCMC and Variational Bayes for Multiway Data

    … on parallelization, geodesic convexity of the evidence lower bound under our metric choice, and flexibility of representation with respect to Monte Carlo simulation. We end this thesis with a third paper, which describes a Bayesian model whose posterior has the capability of interpolating …

    cornell Repository record for Geometric Methods in MCMC and Variational Bayes for Multiway Data (opens in a new tab)

  14. Prospects of Normalizing Flow-Based Differentiable Particle Filters in Target Modelling

    … end-to-end using a combination of a blockwise Evidence Lower Bound (ELBO) loss and the RMSE in position, velocity, and acceleration. The training was conducted on target data generated from stochastic differential equations (SDEs) simulating a range of motion patterns, including turning and …

    chalmers Repository record for Prospects of Normalizing Flow-Based Differentiable Particle Filters in Target Modelling (opens in a new tab)

  15. Tensor decomposition methods and their applications in dynamical systems and generative modeling

    … Appendices provide detailed derivations of the Evidence Lower Bound (ELBO) for variational autoencoders and a mathematical treatment of diffusion and flow models in terms of forward and reverse stochastic differential equations. Across all chapters, this thesis emphasizes how algorithmically …

    texas Repository record for Tensor decomposition methods and their applications in dynamical systems and generative modeling (opens in a new tab)

  16. Joint Data Modeling Using Variational Autoencoders

    … posteriors, can be trained by maximizing an evidence lower bound approximated by Gibbs sampling. I show comparable performance to existing methods on a variety of toy and benchmark datasets in generating realistic samples, with applications to the simultaneous modeling of brain calcium …

    duke Repository record for Joint Data Modeling Using Variational Autoencoders (opens in a new tab)

  17. A BAYESIAN CONVOLUTIONAL NEURAL NETWORK BASED CLASSIFIER TO DETECT BREAST CANCER FROM HISTOPATHOLOGICAL IMAGES AND UNCERTAINTY QUANTIFICATION

    … is known as a Variational Free Energy (VFE) or Evidence Lower Bound (ELBO) which is to be optimized. Since the Bayesian approach provides probability distributions on the weights of the neural networks, it is possible to calculate the variance of the predictive posterior probability …

    mtu Repository record for A BAYESIAN CONVOLUTIONAL NEURAL NETWORK BASED CLASSIFIER TO DETECT BREAST CANCER FROM HISTOPATHOLOGICAL IMAGES AND UNCERTAINTY QUANTIFICATION (opens in a new tab)

  18. Statistical and Computational Properties of Variational Inference

    … an alternative objective function known as the evidence lower bound $\text{ELBO}(q)= \mathbb{E}[\log p(\theta,X^n)] - \mathbb{E}[\log q(\theta)] $ is used to perform this optimization. The following equality shows that minimizing the KL divergence is equivalent to maximizing the ELBO, …

    tamu Repository record for Statistical and Computational Properties of Variational Inference (opens in a new tab)

  19. Energy Efficient Plasticity in Probabilistic Synapses: Expression Mechanisms and Bayesian Principles

    … Gaussian entropy---an essential term in the evidence-lower-bound (ELBO)---optimised in variational inference to approximate Bayesian inference.<br/><br/>We also identified that quantal state expressed through variables: \(n\) (vesicle number), \(p\) (release probability), and \(q\) (quantal …

    bristol Repository record for Energy Efficient Plasticity in Probabilistic Synapses: Expression Mechanisms and Bayesian Principles (opens in a new tab) Download PDF for Energy Efficient Plasticity in Probabilistic Synapses: Expression Mechanisms and Bayesian Principles (opens in a new tab)

  20. Modelling non-stationary functions with Gaussian processes

    … are optimized with respect to the variational evidence lower bound (ELBO). Second, this thesis presents a kernel suitable for modelling non-stationary couplings between multiple output variables of interest in the context of multi-task or multi-output GP regression. The construction of the …

    aalto Repository record for Modelling non-stationary functions with Gaussian processes (opens in a new tab)

Page 1 of 2