Enregistré dans:
Détails bibliographiques
Auteurs principaux: Loosmore, Stan, Titus, Alexander
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2411.13786
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866917844268089344
author Loosmore, Stan
Titus, Alexander
author_facet Loosmore, Stan
Titus, Alexander
contents Modern day Language Models see extensive use in text classification, yet this comes at significant computational cost. Compute-effective classification models are needed for low-resource environments, most notably on edge devices. We introduce Adaptable Embeddings Networks (AEN), a novel dual-encoder architecture using Kernel Density Estimation (KDE). This architecture allows for runtime adaptation of classification criteria without retraining and is non-autoregressive. Through thorough synthetic data experimentation, we demonstrate our model outputs comparable and in certain cases superior results to that of autoregressive models an order of magnitude larger than AEN's size. The architecture's ability to preprocess and cache condition embeddings makes it ideal for edge computing applications and real-time monitoring systems.
format Preprint
id arxiv_https___arxiv_org_abs_2411_13786
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Adaptable Embeddings Network (AEN)
Loosmore, Stan
Titus, Alexander
Machine Learning
Computation and Language
Modern day Language Models see extensive use in text classification, yet this comes at significant computational cost. Compute-effective classification models are needed for low-resource environments, most notably on edge devices. We introduce Adaptable Embeddings Networks (AEN), a novel dual-encoder architecture using Kernel Density Estimation (KDE). This architecture allows for runtime adaptation of classification criteria without retraining and is non-autoregressive. Through thorough synthetic data experimentation, we demonstrate our model outputs comparable and in certain cases superior results to that of autoregressive models an order of magnitude larger than AEN's size. The architecture's ability to preprocess and cache condition embeddings makes it ideal for edge computing applications and real-time monitoring systems.
title Adaptable Embeddings Network (AEN)
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2411.13786