Edge Pruning
Edge pruning is a neural network compression technique aiming to reduce computational costs and memory usage by removing less important connections or parameters without significant performance degradation. Current research focuses on developing efficient pruning algorithms for various architectures, including convolutional neural networks (CNNs), vision transformers (ViTs), and large language models (LLMs), often incorporating techniques like knowledge distillation and optimization-based methods to improve performance after pruning. This work is significant because it enables the deployment of large, powerful models on resource-constrained devices and improves the energy efficiency of training and inference, impacting both scientific understanding of model redundancy and practical applications across diverse fields.
Papers
Towards Communication-Learning Trade-off for Federated Learning at the Network Edge
Jianyang Ren, Wanli Ni, Hui Tian
Prune and distill: similar reformatting of image information along rat visual cortex and deep neural networks
Paolo Muratore, Sina Tafazoli, Eugenio Piasini, Alessandro Laio, Davide Zoccolan
Pruning has a disparate impact on model accuracy
Cuong Tran, Ferdinando Fioretto, Jung-Eun Kim, Rakshit Naidu
FCN-Pose: A Pruned and Quantized CNN for Robot Pose Estimation for Constrained Devices
Marrone Silvério Melo Dantas, Iago Richard Rodrigues, Assis Tiago Oliveira Filho, Gibson Barbosa, Daniel Bezerra, Djamel F. H. Sadok, Judith Kelner, Maria Marquezini, Ricardo Silva