Dynamic Probabilistic Pruning: A General Framework for Hardware-Constrained Pruning at Different Granularities

Lizeth Gonzalez-Carabarin (Corresponding author), Iris A.M. Huijben, Bastian Veeling, Alexandre Schmid, Ruud J.G. van Sloun

Onderzoeksoutput: Bijdrage aan tijdschriftTijdschriftartikelAcademicpeer review

7 Citaten (Scopus)
30 Downloads (Pure)

Samenvatting

Unstructured neural network pruning algorithms have achieved impressive compression ratios. However, the resulting, typically irregular sparse matrices hamper efficient hardware implementations, leading to additional memory usage and complex control logic that diminishes the benefits of unstructured pruning. This has spurred structured coarse-grained pruning solutions that prune entire feature maps or even layers, enabling efficient implementation at the expense of reduced flexibility. Here, we propose a flexible new pruning mechanism that facilitates pruning at different granularities (weights, kernels, and feature maps) while retaining efficient memory organization (e.g., pruning exactly k-out-of-n weights for every output neuron or pruning exactly k-out-of-n kernels for every feature map). We refer to this algorithm as dynamic probabilistic pruning (DPP). DPP leverages the Gumbel-softmax relaxation for differentiable k-out-of-n sampling, facilitating end-to-end optimization. We show that DPP achieves competitive compression ratios and classification accuracy when pruning common deep learning models trained on different benchmark datasets for image classification. Relevantly, the dynamic masking of DPP facilitates for joint optimization of pruning and weight quantization in order to even further compress the network, which we show as well. Finally, we propose novel information-theoretic metrics that show the confidence and pruning diversity of pruning masks within a layer.

Originele taal-2Engels
Artikelnummer9790881
Pagina's (van-tot)733-744
Aantal pagina's12
TijdschriftIEEE Transactions on Neural Networks and Learning Systems
Volume35
Nummer van het tijdschrift1
Vroegere onlinedatum8 jun. 2022
DOI's
StatusGepubliceerd - 1 jan. 2024

Financiering

This work was supported by the Marie Sklodowska-Curie Actions.

FinanciersFinanciernummer
H2020 Marie Skłodowska-Curie Actions

    Vingerafdruk

    Duik in de onderzoeksthema's van 'Dynamic Probabilistic Pruning: A General Framework for Hardware-Constrained Pruning at Different Granularities'. Samen vormen ze een unieke vingerafdruk.

    Citeer dit