Salvato in:
Dettagli Bibliografici
Autori principali: C, Shiva Kumar, Dhiman, Jitendra Kumar, Adiga, Nagaraj, Singh, Shatrughan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2502.05837
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912225797603328
author C, Shiva Kumar
Dhiman, Jitendra Kumar
Adiga, Nagaraj
Singh, Shatrughan
author_facet C, Shiva Kumar
Dhiman, Jitendra Kumar
Adiga, Nagaraj
Singh, Shatrughan
contents Traditionally, Knowledge Distillation (KD) is used for model compression, often leading to suboptimal performance. In this paper, we evaluate the impact of combining KD loss with alternative pruning techniques, including Low-Rank Factorization (LRF) and l0 regularization, on a conformer-based pre-trained network under the paradigm of Self-Supervised Learning (SSL). We also propose a strategy to jointly prune and train an RNN-T-based ASR model, demonstrating that this approach yields superior performance compared to pruning a pre-trained network first and then using it for ASR training. This approach led to a significant reduction in word error rate: l0 and KD combination achieves the best non-streaming performance, with a 8.9% Relative Word Error Rate (RWER) improvement over the baseline, while LRF and KD combination yields the best results for streaming ASR, improving RWER by 13.4%.
format Preprint
id arxiv_https___arxiv_org_abs_2502_05837
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Synergistic Effects of Knowledge Distillation and Structured Pruning for Self-Supervised Speech Models
C, Shiva Kumar
Dhiman, Jitendra Kumar
Adiga, Nagaraj
Singh, Shatrughan
Audio and Speech Processing
Traditionally, Knowledge Distillation (KD) is used for model compression, often leading to suboptimal performance. In this paper, we evaluate the impact of combining KD loss with alternative pruning techniques, including Low-Rank Factorization (LRF) and l0 regularization, on a conformer-based pre-trained network under the paradigm of Self-Supervised Learning (SSL). We also propose a strategy to jointly prune and train an RNN-T-based ASR model, demonstrating that this approach yields superior performance compared to pruning a pre-trained network first and then using it for ASR training. This approach led to a significant reduction in word error rate: l0 and KD combination achieves the best non-streaming performance, with a 8.9% Relative Word Error Rate (RWER) improvement over the baseline, while LRF and KD combination yields the best results for streaming ASR, improving RWER by 13.4%.
title Synergistic Effects of Knowledge Distillation and Structured Pruning for Self-Supervised Speech Models
topic Audio and Speech Processing
url https://arxiv.org/abs/2502.05837