S2WTM: Spherical Sliced-Wasserstein Autoencoder for Topic Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Adhya, Suman, Sanyal, Debarshi Kumar
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909737366323200
author Adhya, Suman
Sanyal, Debarshi Kumar
author_facet Adhya, Suman
Sanyal, Debarshi Kumar
contents Modeling latent representations in a hyperspherical space has proven effective for capturing directional similarities in high-dimensional text data, benefiting topic modeling. Variational autoencoder-based neural topic models (VAE-NTMs) commonly adopt the von Mises-Fisher prior to encode hyperspherical structure. However, VAE-NTMs often suffer from posterior collapse, where the KL divergence term in the objective function highly diminishes, leading to ineffective latent representations. To mitigate this issue while modeling hyperspherical structure in the latent space, we propose the Spherical Sliced Wasserstein Autoencoder for Topic Modeling (S2WTM). S2WTM employs a prior distribution supported on the unit hypersphere and leverages the Spherical Sliced-Wasserstein distance to align the aggregated posterior distribution with the prior. Experimental results demonstrate that S2WTM outperforms state-of-the-art topic models, generating more coherent and diverse topics while improving performance on downstream tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2507_12451
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle S2WTM: Spherical Sliced-Wasserstein Autoencoder for Topic Modeling
Adhya, Suman
Sanyal, Debarshi Kumar
Computation and Language
Artificial Intelligence
Machine Learning
Modeling latent representations in a hyperspherical space has proven effective for capturing directional similarities in high-dimensional text data, benefiting topic modeling. Variational autoencoder-based neural topic models (VAE-NTMs) commonly adopt the von Mises-Fisher prior to encode hyperspherical structure. However, VAE-NTMs often suffer from posterior collapse, where the KL divergence term in the objective function highly diminishes, leading to ineffective latent representations. To mitigate this issue while modeling hyperspherical structure in the latent space, we propose the Spherical Sliced Wasserstein Autoencoder for Topic Modeling (S2WTM). S2WTM employs a prior distribution supported on the unit hypersphere and leverages the Spherical Sliced-Wasserstein distance to align the aggregated posterior distribution with the prior. Experimental results demonstrate that S2WTM outperforms state-of-the-art topic models, generating more coherent and diverse topics while improving performance on downstream tasks.
title S2WTM: Spherical Sliced-Wasserstein Autoencoder for Topic Modeling
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2507.12451