Sparsification via Compressed Sensing for Automatic Speech Recognition
This work aims to improve the efficiency of ASR models for real-time applications on resource-constrained devices, which is a significant problem for users requiring low-latency interactions. It offers an incremental improvement over existing sparse pruning techniques.
The paper addresses the challenge of reducing model size and latency in Automatic Speech Recognition (ASR) by proposing a compressed sensing based pruning (CSP) approach. This method reformulates sparse pruning as a dual problem of sparsity induction and compression-error reduction, integrating compressed sensing into the machine learning training process. The authors claim that CSP consistently outperforms existing pruning methods for ASR.
In order to achieve high accuracy for machine learning (ML) applications, it is essential to employ models with a large number of parameters. Certain applications, such as Automatic Speech Recognition (ASR), however, require real-time interactions with users, hence compelling the model to have as low latency as possible. Deploying large scale ML applications thus necessitates model quantization and compression, especially when running ML models on resource constrained devices. For example, by forcing some of the model weight values into zero, it is possible to apply zero-weight compression, which reduces both the model size and model reading time from the memory. In the literature, such methods are referred to as sparse pruning. The fundamental questions are when and which weights should be forced to zero, i.e. be pruned. In this work, we propose a compressed sensing based pruning (CSP) approach to effectively address those questions. By reformulating sparse pruning as a sparsity inducing and compression-error reduction dual problem, we introduce the classic compressed sensing process into the ML model training process. Using ASR task as an example, we show that CSP consistently outperforms existing approaches in the literature.