Task-Agnostic Structured Pruning of Speech Representation Models

被引:1
|
作者
Wang, Haoyu [1 ]
Wang, Siyuan [1 ]
Zhang, Wei-Qiang [1 ]
Suo, Hongbin [2 ]
Wan, Yulong [2 ]
机构
[1] Tsinghua Univ, Dept Elect Engn, Beijing 100084, Peoples R China
[2] OPPO, Data & AI Engn Syst, Beijing 100026, Peoples R China
来源
基金
中国国家自然科学基金;
关键词
Model pruning; knowledge distillation; model compression; representation learning;
D O I
10.21437/Interspeech.2023-1442
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
Self-supervised pre-trained models such as Wav2vec2, Hubert, and WavLM have been shown to significantly improve many speech tasks. However, their large memory and strong computational requirements hinder their industrial applicability. Structured pruning is a hardware-friendly model compression technique but usually results in a larger loss of accuracy. In this paper, we propose a fine-grained attention head pruning method to compensate for the performance degradation. In addition, we also introduce the straight through estimator into the L0 regularization to further accelerate the pruned model. Experiments on the SUPERB benchmark show that our model can achieve comparable performance to the dense model in multiple tasks and outperforms the Wav2vec 2.0 base model on average, with 72% fewer parameters and 2 times faster inference speed.
引用
收藏
页码:231 / 235
页数:5
相关论文
共 50 条
  • [1] Hierarchically structured task-agnostic continual learning
    Heinke Hihn
    Daniel A. Braun
    Machine Learning, 2023, 112 : 655 - 686
  • [2] Hierarchically structured task-agnostic continual learning
    Hihn, Heinke
    Braun, Daniel A.
    MACHINE LEARNING, 2023, 112 (02) : 655 - 686
  • [3] MULTIFLOW: Shifting Towards Task-Agnostic Vision-Language Pruning
    Farina, Matteo
    Mancini, Massimiliano
    Cunegatti, Elia
    Liu, Gaowen
    Iacca, Giovanni
    Ricci, Elisa
    2024 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2024, : 16185 - 16195
  • [4] Unsupervised Representation Learning with Task-Agnostic Feature Masking for Robust End-to-End Speech Recognition
    Kim, June-Woo
    Chung, Hoon
    Jung, Ho-Young
    MATHEMATICS, 2023, 11 (03)
  • [5] Task-Agnostic Graph Explanations
    Xie, Yaochen
    Katariya, Sumeet
    Tang, Xianfeng
    Huang, Edward
    Rao, Nikhil
    Subbian, Karthik
    Ji, Shuiwang
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022), 2022,
  • [6] Task-agnostic representation learning of multimodal twitter data for downstream applications
    Ryan Rivas
    Sudipta Paul
    Vagelis Hristidis
    Evangelos E. Papalexakis
    Amit K. Roy-Chowdhury
    Journal of Big Data, 9
  • [7] TRIO: Task-agnostic dataset representation optimized for automatic algorithm selection
    Cohen-Shapira, Noy
    Rokach, Lior
    2021 21ST IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2021), 2021, : 81 - 90
  • [8] Task-agnostic representation learning of multimodal twitter data for downstream applications
    Rivas, Ryan
    Paul, Sudipta
    Hristidis, Vagelis
    Papalexakis, Evangelos E.
    Roy-Chowdhury, Amit K.
    JOURNAL OF BIG DATA, 2022, 9 (01)
  • [9] Fundamentals of Task-Agnostic Data Valuation
    Amiri, Mohammad Mohammadi
    Berdoz, Frederic
    Raskar, Ramesh
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 8, 2023, : 9226 - 9234
  • [10] Task-Agnostic Safety for Reinforcement Learning
    Rahman, Md Asifur
    Alqahtani, Sarra
    PROCEEDINGS OF THE 16TH ACM WORKSHOP ON ARTIFICIAL INTELLIGENCE AND SECURITY, AISEC 2023, 2023, : 139 - 148