Linear Complexity Self-Supervised Learning for Music Understanding with Random Quantizer

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Vavaroutsos, Petros, Palamas, Theodoros, Vikatos, Pantelis
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911376265445376
author Vavaroutsos, Petros
Palamas, Theodoros
Vikatos, Pantelis
author_facet Vavaroutsos, Petros
Palamas, Theodoros
Vikatos, Pantelis
contents In recent years, foundation models have become very popular due to their exceptional performance, mainly in natural language (NLP) tasks where they were first introduced. These models usually consist of hundreds of millions, or even billions, of parameters, making them resource-intensive during training and in production systems, leading to increased costs. This paper focuses on the reduction of a foundation's model size when applied to music information retrieval (MIR) tasks. Our research combines the Branchformer architecture with SummaryMixing, which were first applied in speech recognition, along with a random quantization process. To facilitate reproducibility, we conduct pre-training on publicly available datasets, complemented by a proprietary dataset comparable in scale to other private datasets reported in the literature. We ensure robust evaluation by using a framework consisting of a variety of downstream MIR tasks. Our results show that our architecture achieves competitive performance when compared with other state-of-the-art models that use multi-head self-attention, while reducing the model size from 8.5% up to 12.3%.
format Preprint
id arxiv_https___arxiv_org_abs_2601_09603
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Linear Complexity Self-Supervised Learning for Music Understanding with Random Quantizer
Vavaroutsos, Petros
Palamas, Theodoros
Vikatos, Pantelis
Sound
Artificial Intelligence
Computation and Language
Machine Learning
In recent years, foundation models have become very popular due to their exceptional performance, mainly in natural language (NLP) tasks where they were first introduced. These models usually consist of hundreds of millions, or even billions, of parameters, making them resource-intensive during training and in production systems, leading to increased costs. This paper focuses on the reduction of a foundation's model size when applied to music information retrieval (MIR) tasks. Our research combines the Branchformer architecture with SummaryMixing, which were first applied in speech recognition, along with a random quantization process. To facilitate reproducibility, we conduct pre-training on publicly available datasets, complemented by a proprietary dataset comparable in scale to other private datasets reported in the literature. We ensure robust evaluation by using a framework consisting of a variety of downstream MIR tasks. Our results show that our architecture achieves competitive performance when compared with other state-of-the-art models that use multi-head self-attention, while reducing the model size from 8.5% up to 12.3%.
title Linear Complexity Self-Supervised Learning for Music Understanding with Random Quantizer
topic Sound
Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2601.09603