Skip to content

intsystems/BHPO-AE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Selecting-the-Optimal-Autoencoder-Structure-Using-Bayesian-Optimization

Overview

This repository explores the task of selecting an optimal autoencoder architecture using Bayesian optimization. An autoencoder is a differentiable model composed of two main components:

  • Encoder: Transforms the input data into a latent vector representation.
  • Decoder: Reconstructs the original input from the latent representation.

The structure of the autoencoder is defined by a set of hyperparameters, which are optimized using a novel two-stage Bayesian optimization approach.

Proposed Method

We introduce a modified Bayesian optimization technique with the following key steps:

  1. Candidate Selection: At each iteration, a set of points with the best model quality estimates is chosen.
  2. Dynamic Learning Analysis: The best candidate is selected based on training dynamics and performance.

The approach is theoretically justified and experimentally validated on the CIFAR and Fashion-MNIST datasets, demonstrating improved efficiency in architecture selection.

Key Features

  • Bayesian Optimization: Efficient hyperparameter tuning for autoencoder architectures.
  • Two-Stage Selection: Combines quality estimation with training dynamics for better convergence.
  • Theoretical & Empirical Validation: Supported by both theoretical analysis and experiments.

Datasets

  • CIFAR
  • Fashion-MNIST

About

selecting an optimal autoencoder architecture using Bayesian optimization

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published