pruning strategy
Recently Published Documents


TOTAL DOCUMENTS

104
(FIVE YEARS 45)

H-INDEX

11
(FIVE YEARS 3)

2021 ◽  
Author(s):  
Yu-Pei Liang ◽  
Yung-Han Hsu ◽  
Tseng-Yi Chen ◽  
Shuo-Han Chen ◽  
Hsin-Wen Wei ◽  
...  

Author(s):  
Yanir González-Díaz ◽  
José Fco. Martínez-Trinidad ◽  
Jesús A. Carrasco-Ochoa ◽  
Manuel S. Lazo-Cortés
Keyword(s):  

2021 ◽  
Vol 14 (13) ◽  
pp. 3322-3334
Author(s):  
Yunkai Lou ◽  
Chaokun Wang ◽  
Tiankai Gu ◽  
Hao Feng ◽  
Jun Chen ◽  
...  

Many real-world networks have been evolving, and are finely modeled as temporal graphs from the viewpoint of the graph theory. A temporal graph is informative, and always contains two types of information, i.e., the temporal information and topological information, where the temporal information reflects the time when the relationships are established, and the topological information focuses on the structure of the graph. In this paper, we perform time-topology analysis on temporal graphs to extract useful information. Firstly, a new metric named T-cohesiveness is proposed to evaluate the cohesiveness of a temporal subgraph. It defines the cohesiveness of a temporal subgraph from the time and topology dimensions jointly. Specifically, given a temporal graph G s = ( Vs , ε Es ), cohesiveness in the time dimension reflects whether the connections in G s happen in a short period of time, while cohesiveness in the topology dimension indicates whether the vertices in V s are densely connected and have few connections with vertices out of G s . Then, T-cohesiveness is utilized to perform time-topology analysis on temporal graphs, and two time-topology analysis methods are proposed. In detail, T-cohesiveness evolution tracking traces the evolution of the T-cohesiveness of a subgraph, and combo searching finds out all the subgraphs that contain the query vertex and have T-cohesiveness larger than a given threshold. Moreover, a pruning strategy is proposed to improve the efficiency of combo searching. Experimental results confirm the efficiency of the proposed time-topology analysis methods and the pruning strategy.


2021 ◽  
pp. 1-14
Author(s):  
Heng Wang ◽  
Xiang Ye ◽  
Yong Li

Model pruning aims to reduce the parameter amount of deep neural networks while retaining the performance. Existing strategies often treat all layers equally and all layers simply share the same pruning rate. However, it is observed from our experiments that the redundancy degree differs from layer to layer. Based on this observation, this work proposes a pruning strategy depending on the layer-wise redundancy degree. Firstly, we define the redundancy degree for each layer by the norm and similarity redundancy of filters. Then a novel layer-wise strategy, Redundancy-dependent Filter Pruning (RedFiP), is proposed which prunes different proportion of filters at different layers according to the defined redundancy degree. Since the redundancy analysis and experimental results of RedFiP show that deeper layers need fewer filters, a phase-wise strategy, Phased Filter Pruning (PFP), is proposed that divides the layers into three phases and layers in each phase share the same pruning rate. The phase-wise PFP allows the layer-wise RedFiP to be easily implemented in existing structures of deep neural networks. Experimental results show that when total parameters are pruned by 40%, RedFiP outperforms the state-of-the-art strategy FPGM-Mixed by 1.83% on CIFAR-100, and even slightly outperforms the non-pruned model by 0.11% on CIFAR-10. On ImageNet-1k, RedFiP (30%) and PFP (30%) outperform FPGM-Mixed (30%) by 1.3% and 0.8% with ResNet-18.


Entropy ◽  
2021 ◽  
Vol 23 (8) ◽  
pp. 1042
Author(s):  
Lan Huang ◽  
Jia Zeng ◽  
Shiqi Sun ◽  
Wencong Wang ◽  
Yan Wang ◽  
...  

Deep neural networks may achieve excellent performance in many research fields. However, many deep neural network models are over-parameterized. The computation of weight matrices often consumes a lot of time, which requires plenty of computing resources. In order to solve these problems, a novel block-based division method and a special coarse-grained block pruning strategy are proposed in this paper to simplify and compress the fully connected structure, and the pruned weight matrices with a blocky structure are then stored in the format of Block Sparse Row (BSR) to accelerate the calculation of the weight matrices. First, the weight matrices are divided into square sub-blocks based on spatial aggregation. Second, a coarse-grained block pruning procedure is utilized to scale down the model parameters. Finally, the BSR storage format, which is much more friendly to block sparse matrix storage and computation, is employed to store these pruned dense weight blocks to speed up the calculation. In the following experiments on MNIST and Fashion-MNIST datasets, the trend of accuracies with different pruning granularities and different sparsity is explored in order to analyze our method. The experimental results show that our coarse-grained block pruning method can compress the network and can reduce the computational cost without greatly degrading the classification accuracy. The experiment on the CIFAR-10 dataset shows that our block pruning strategy can combine well with the convolutional networks.


2021 ◽  
pp. 1-22
Author(s):  
Haodong Cheng ◽  
Meng Han ◽  
Ni Zhang ◽  
Le Wang ◽  
Xiaojuan Li

The researcher proposed the concept of Top-K high-utility itemsets mining over data streams. Users directly specify the number K of high-utility itemsets they wish to obtain for mining with no need to set a minimum utility threshold. There exist some problems in current Top-K high-utility itemsets mining algorithms over data streams including the complex construction process of the storage structure, the inefficiency of threshold raising strategies and utility pruning strategies, and large scale of the search space, etc., which still can not meet the requirement of real-time processing over data streams with limited time and memory constraints. To solve this problem, this paper proposes an efficient algorithm based on dataset projection for mining Top-K high-utility itemsets from a data stream. A data structure CIUDataListSW is also proposed, which stores the position of the item in the transaction to effectively obtain the initial projected dataset of the item. In order to improve the projection efficiency, this paper innovates a new reorganization technology for projected transactions in common batches to maintain the sort order of transactions in the process of dataset projection. Dual pruning strategy and transaction merging mechanism are also used to further reduce search space and dataset scanning costs. In addition, based on the proposed CUDH S W structure, an efficient threshold raising strategy CUD is used, and a new threshold raising strategy CUDCB is designed to further shorten the mining time. Experimental results show that the algorithm has great advantages in running time and memory consumption, and it is especially suitable for the mining of high-utility itemsets of dense datasets.


2021 ◽  
pp. 1-13
Author(s):  
Xinghao Chen ◽  
Bin Zhou

Path planning is the basis and prerequisite for unmanned aerial vehicle (UAV) to perform tasks, and it is important to achieve precise location in path planning. This paper focuses on solving the UAV path planning problem under the constraint of system positioning error. Some nodes can re-initiate the accumulated flight error to zero and this type of scenario can be modeled as the resource-constrained shortest path problem with re-initialization (RCSPP-R). The additional re-initiation conditions expand the set of viable paths for the original constrained shortest path problem and increasing the search cost. To solve the problem, an effective preprocessing method is proposed to reduce the network nodes. At the same time, a relaxed pruning strategy is introduced into the traditional Pulse algorithm to reduce the search space and avoid more redundant calculations on unfavorable scalable nodes by the proposed heuristic search strategy. To evaluate the accuracy and effectiveness of the proposed algorithm, some numerical experiments were carried out. The results indicate that the three strategies can reduce the search space by 99%, 97% and 80%, respectively, and in the case of a large network, the heuristic algorithm combining the three strategies can improve the efficiency by an average of 80% compared to some classical solution.


Electronics ◽  
2021 ◽  
Vol 10 (15) ◽  
pp. 1830
Author(s):  
Jiabao Gao ◽  
Qingliang Liu ◽  
Jinmei Lai

Binarized neural networks (BNNs), which have 1-bit weights and activations, are well suited for FPGA accelerators as their dominant computations are bitwise arithmetic, and the reduction in memory requirements means that all the network parameters can be stored in internal memory. However, the energy efficiency of these accelerators is still restricted by the abundant redundancies in BNNs. This hinders their deployment for applications in smart sensors and tiny devices because these scenarios have tight constraints with respect to energy consumption. To overcome this problem, we propose an approach to implement BNN inference while offering excellent energy efficiency for the accelerators by means of pruning the massive redundant operations while maintaining the original accuracy of the networks. Firstly, inspired by the observation that the convolution processes of two related kernels contain many repeated computations, we first build one formula to clarify the reusing relationships between their convolutional outputs and remove the unnecessary operations. Furthermore, by generalizing this reusing relationship to one tile of kernels in one neuron, we adopt an inclusion pruning strategy to further skip the superfluous evaluations of the neurons whose real output values can be determined early. Finally, we evaluate our system on the Zynq 7000 XC7Z100 FPGA platform. Our design can prune 51 percent of the operations without any accuracy loss. Meanwhile, the energy efficiency of our system is as high as 6.55 × 105 Img/kJ, which is 118× better than the best accelerator based on an NVDIA Tesla-V100 GPU and 3.6× higher than the state-of-the-art FPGA implementations for BNNs.


Sign in / Sign up

Export Citation Format

Share Document