Local Disentanglement in Variational Auto-Encoders Using Jacobian $L_1$ Regularization

2021-06-05 · NeurIPS 2021 12 · Travers Rhodes, Daniel D. Lee

There have been many recent advances in representation learning; however, unsupervised representation learning can still struggle with model identification issues related to rotations of the latent space. Variational Auto-Encoders (VAEs) and their extensions such as $\beta$-VAEs have been shown to improve local alignment of latent variables with PCA directions, which can help to improve model disentanglement under some conditions. Borrowing inspiration from Independent Component Analysis (ICA) and sparse coding, we propose applying an $L_1$ loss to the VAE's generative Jacobian during training to encourage local latent variable alignment with independent factors of variation in images of multiple objects or images with multiple parts. We demonstrate our results on a variety of datasets, giving qualitative and quantitative results using information theoretic and modularity measures that show our added $L_1$ cost encourages local axis alignment of the latent representation with individual factors of variation.

📄 PDF Abstract BibTeX arXiv:2106.02923

Code (1)

travers-rhodes/jlonevae 공식 구현 tf

Tasks

Disentanglement Representation Learning

Methods 이 논문이 사용한 방법론

PCA Principle Components Analysis (PCA) is an unsupervised method primary used for dimensionality reduction within machine learning. PCA is calculated via a singular value…

Unpicking Data at the Seams: VAEs, Disentanglement and Independent Components

2024-10-29 · Carl Allen

Disentanglement, or identifying salient statistically independent factors of the data, is of interest in many areas of machine learning and statistics, such as synthetic data generation with controlled properties, robust…

Disentanglement Robust classification Synthetic Data Generation

Disentanglement Learning for Variational Autoencoders Applied to Audio-Visual Speech Enhancement

2021-05-19 · Guillaume Carbajal, Julius Richter, Timo Gerkmann

Recently, the standard variational autoencoder has been successfully used to learn a probabilistic prior over speech signals, which is then used to perform speech enhancement. Variational autoencoders have then been cond…

Attribute Decoder Disentanglement Speech Enhancement

Quantifying the Effects of Enforcing Disentanglement on Variational Autoencoders

2017-11-24 · Momchil Peychev, Petar Veličković, Pietro Liò

The notion of disentangled autoencoders was proposed as an extension to the variational autoencoder by introducing a disentanglement parameter $\beta$, controlling the learning pressure put on the possible underlying lat…

Disentanglement

On Implicit Regularization in $β$-VAEs

2020-01-31 · Abhishek Kumar, Ben Poole

While the impact of variational inference (VI) on posterior inference in a fixed generative model is well-characterized, its role in regularizing a learned generative model when used in variational autoencoders (VAEs) is…

Variational Inference

On Implicit Regularization in $\beta$-VAEs

2020-01-01 · ICML 2020 1 · Abhishek Kumar, Ben Poole