Redirigiendo al acceso original de articulo en 19 segundos...
Inicio  /  Information  /  Vol: 12 Par: 7 (2021)  /  Artículo
ARTÍCULO
TITULO

Combine-Net: An Improved Filter Pruning Algorithm

Jinghan Wang    
Guangyue Li and Wenzhao Zhang    

Resumen

The powerful performance of deep learning is evident to all. With the deepening of research, neural networks have become more complex and not easily generalized to resource-constrained devices. The emergence of a series of model compression algorithms makes artificial intelligence on edge possible. Among them, structured model pruning is widely utilized because of its versatility. Structured pruning prunes the neural network itself and discards some relatively unimportant structures to compress the model?s size. However, in the previous pruning work, problems such as evaluation errors of networks, empirical determination of pruning rate, and low retraining efficiency remain. Therefore, we propose an accurate, objective, and efficient pruning algorithm?Combine-Net, introducing Adaptive BN to eliminate evaluation errors, the Kneedle algorithm to determine the pruning rate objectively, and knowledge distillation to improve the efficiency of retraining. Results show that, without precision loss, Combine-Net achieves 95% parameter compression and 83% computation compression on VGG16 on CIFAR10, 71% of parameter compression and 41% computation compression on ResNet50 on CIFAR100. Experiments on different datasets and models have proved that Combine-Net can efficiently compress the neural network?s parameters and computation.

 Artículos similares

       
 
Zhihua Li, Lin Zhang and Kunlun Wu    
The laser gyroscope of Laser Inertial Navigation System (LINS) eliminates the influence of the locked zone with mechanical dither. The output information of laser gyroscopes must be filtered before use to eliminate vibration noise. Laser gyroscope filter... ver más
Revista: Aerospace

 
Lichao Nengzi, Ying Jiang, Zhirong Fang, Qiyuan Hu, Guanglei Qiu and Haitao Li    
In order to improve the removal efficiency of refractory organic matters in micro-polluted source water, biological manganese oxides (BMOs) were generated in situ in the biological aerated filter (BAF) (BAF 2#), which could oxidize the refractory organic... ver más
Revista: Water

 
Xuejia Wu, Zhong Su, Lei Li and Zekun Bai    
To address the issue of low positioning accuracy in unmanned vehicles navigating in obstructed spaces due to easily contaminated navigation measurement information, an improved adaptive federated Kalman filtering INS/GNSS/VNS integrated navigation algori... ver más
Revista: Applied Sciences

 
Xiaoyu Liu and Xiang Zhang    
In order to enhance the accuracy of the traditional extended Kalman filter (EKF) algorithm in the estimation of the state of charge (SoC) of power batteries, we first derived the state space equation and measurement equation of lithium power batteries ba... ver más
Revista: Applied Sciences

 
Binghui Tang, Yuemin Wang, Ang Chen, Ruqing Gong and Yunwei Zhao    
Pipeline systems are prone to defects due to the harsh service conditions, which may induce catastrophic failure if found not in time. Ultrasonic guided wave (UGW) testing provides a convenient option for pipeline detection, showing high-efficiency, non-... ver más
Revista: Applied Sciences