List of Articles pruning

• Open Access Article
  • Abstract Page
  • Full-Text
  
  1 - Speed up the Search for Proximity-Based Models
  J. Paksima A. Zareh V. Derhami
  One of the main challenges in the proximity models is the speed of data retrieval. These models define a distance concept which is calculated based on the positions of query terms in the documents. This means that finding the positions and calculating the distance is a More

  One of the main challenges in the proximity models is the speed of data retrieval. These models define a distance concept which is calculated based on the positions of query terms in the documents. This means that finding the positions and calculating the distance is a time consuming process and because it usually executed during the search time it has a special importance to users. If we can reduce the number of documents, retrieval process becomes faster. In this paper, the SNTK3 algorithm is proposed to prune documents dynamically. To avoid allocating too much memory and reducing the risk of errors during the retrieval, some documents' scores are calculated without any pruning (Skip-N). The SNTK3 algorithm uses three pyramids to extract documents with the highest scores. Experiments show that the proposed algorithm can improve the speed of retrieval. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  2 - Multi-level ternary quantization for improving sparsity and computation in embedded deep neural networks
  Hosna Manavi Mofrad Seyed Ali ansarmohammadi Mostafa Salehi
  Deep neural networks (DNNs) have achieved great interest due to their success in various applications. However, the computation complexity and memory size are considered to be the main obstacles for implementing such models on embedded devices with limited memory and co More

  Deep neural networks (DNNs) have achieved great interest due to their success in various applications. However, the computation complexity and memory size are considered to be the main obstacles for implementing such models on embedded devices with limited memory and computational resources. Network compression techniques can overcome these challenges. Quantization and pruning methods are the most important compression techniques among them. One of the famous quantization methods in DNNs is the multi-level binary quantization, which not only exploits simple bit-wise logical operations, but also reduces the accuracy gap between binary neural networks and full precision DNNs. Since, multi-level binary can’t represent the zero value, this quantization does’nt take advantage of sparsity. On the other hand, it has been shown that DNNs are sparse, and by pruning the parameters of the DNNs, the amount of data storage in memory is reduced while computation speedup is also achieved. In this paper, we propose a pruning and quantization-aware training method for multi-level ternary quantization that takes advantage of both multi-level quantization and data sparsity. In addition to increasing the accuracy of the network compared to the binary multi-level networks, it gives the network the ability to be sparse. To save memory size and computation complexity, we increase the sparsity in the quantized network by pruning until the accuracy loss is negligible. The results show that the potential speedup of computation for our model at the bit and word-level sparsity can be increased by 15x and 45x compared to the basic multi-level binary networks. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  3 - Analysis of the War in "Haras (The Pruning)" According to the Reception Theory
  Nikou Ghassemi Esfahani Elmira Dadvar
  Reception theory, places a premium on and holds the role of reader in interpreting the text in high regard. Wolfgang Iser is one of the founders of this theory, which believes a text has two artistic and aesthetic poles and relegates the role of meaning creator to the r More

  Reception theory, places a premium on and holds the role of reader in interpreting the text in high regard. Wolfgang Iser is one of the founders of this theory, which believes a text has two artistic and aesthetic poles and relegates the role of meaning creator to the reader. The texts classified as War literature have always had their audiences in the Persian contemporary literature. By having a critical approach to the Iran-Iraq war, different content and a new structure, "Haras (The Pruning)"of NasimMar’ashi, has managed to change the horizon of expectation of Iranian readers concerning War literature. In this paper, the two artistic and aesthetic poles of this novel will be analyzed based on Iser’s views on the reception theory and then we will refer to the elements that have influenced the interaction between readers and the text, leading to a new perception of war. Furthermore, this article attempts to analyze how this text has managed to satisfy the curiosity of contemporary readers by using modern and original elements. Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  4 - Multi-Level Ternary Quantization for Improving Sparsity and Computation in Embedded Deep Neural Networks
  Hosna Manavi Mofrad ali ansarmohammadi Mostafa Salehi
  Deep neural networks (DNNs) have achieved great interest due to their success in various applications. However, the computation complexity and memory size are considered to be the main obstacles for implementing such models on embedded devices with limited memory and co More

  Deep neural networks (DNNs) have achieved great interest due to their success in various applications. However, the computation complexity and memory size are considered to be the main obstacles for implementing such models on embedded devices with limited memory and computational resources. Network compression techniques can overcome these challenges. Quantization and pruning methods are the most important compression techniques among them. One of the famous quantization methods in DNNs is the multi-level binary quantization, which not only exploits simple bit-wise logical operations, but also reduces the accuracy gap between binary neural networks and full precision DNNs. Since, multi-level binary can’t represent the zero value, this quantization does not take advantage of sparsity. On the other hand, it has been shown that DNNs are sparse, and by pruning the parameters of the DNNs, the amount of data storage in memory is reduced while computation speedup is also achieved. In this paper, we propose a pruning and quantization-aware training method for multi-level ternary quantization that takes advantage of both multi-level quantization and data sparsity. In addition to increasing the accuracy of the network compared to the binary multi-level networks, it gives the network the ability to be sparse. To save memory size and computation complexity, we increase the sparsity in the quantized network by pruning until the accuracy loss is negligible. The results show that the potential speedup of computation for our model at the bit and word-level sparsity can be increased by 15x and 45x compared to the basic multi-level binary networks. Manuscript profile