Separate, Dynamic and Differentiable (SMART) Pruner for Block/Output Channel Pruning on Computer Vision Tasks
This addresses the need for accurate, controllable, and stable pruning methods for industrial applications on edge devices, representing a novel integration rather than an incremental improvement.
The paper tackled the problem of block pruning for neural networks on edge devices by introducing the SMART algorithm, which achieved state-of-the-art performance across seven models, four datasets, and three computer vision tasks.
Block pruning, which eliminates contiguous blocks of weights, is a structural pruning method that can significantly enhance the performance of neural processing units (NPUs). In industrial applications, an ideal block pruning algorithm should meet three key requirements: (1) maintain high accuracy across diverse models and tasks, as machine learning deployments on edge devices are typically accuracy-critical; (2) offer precise control over resource constraints to facilitate user adoption; and (3) provide convergence guarantees to prevent performance instability. However, to the best of our knowledge, no existing block pruning algorithm satisfies all three requirements simultaneously. In this paper, we introduce SMART (Separate, Dynamic, and Differentiable) pruning, a novel algorithm designed to address this gap. SMART leverages both weight and activation information to enhance accuracy, employs a differentiable top-k operator for precise control of resource constraints, and offers convergence guarantees under mild conditions. Extensive experiments involving seven models, four datasets, three different block types, and three computer vision tasks demonstrate that SMART pruning achieves state-of-the-art performance in block pruning.