scholarly journals Improved Dynamic Routing Algorithm for Information Aggregation

2021 ◽  
Vol 5 (1) ◽  
Author(s):  
Gongbin Chen ◽  
◽  
Wei Xiang ◽  
Yansong Deng ◽  
◽  
...  
Keyword(s):  
Routing Algorithm ◽  
Information Aggregation ◽  
Dynamic Routing ◽  
Attention Mechanism ◽  
Aggregation Method ◽  
Text Encoding ◽  
Natural Logarithm ◽  
Aggregate Information ◽  
Routing Policy ◽  
Series Of Experiments

Information aggregation is an essential component of text encoding, but it has been paid less attention. The pooling-based (max or average pooling) aggregation method is a bottom-up and passive aggregation method, and loses a lot of important information. Recently, attention mechanism and dynamic routing policy are separately used to aggregate information, but their aggregation capabilities can be further improved. In this paper, we proposed an novel aggregation method combining attention mechanism and dynamic routing, which can strengthen the ability of information aggregation and improve the quality of text encoding. Then, a novel Leaky Natural Logarithm (LNL) squash function is designed to alleviate the “saturation” problem of the squash function of the original dynamic routing. Layer Normalization is added to the dynamic routing policy for speeding up routing convergence as well. A series of experiments are conducted on five text classification benchmarks. Experimental results show that our method outperforms other aggregating methods.

Negative-supervised capsule graph neural network for few-shot text classification

2021 ◽  
pp. 1-13
Author(s):  
Ling Ding ◽  
Xiaojun Chen ◽  
Yang Xiang
Keyword(s):  
Neural Network ◽  
Text Classification ◽  
Routing Algorithm ◽  
Information Aggregation ◽  
Dynamic Routing ◽  
Experimental Results ◽  
Text Data ◽  
Support Set ◽  
Different Types ◽  
Benchmark Datasets

Few-shot text classification aims to learn a classifier from very few labeled text data. Existing studies on this topic mainly adopt prototypical networks and focus on interactive information between support set and query instances to learn generalized class prototypes. However, in the process of encoding, these methods only pay attention to the matching information between support set and query instances, and ignore much useful information about intra-class similarity and inter-class dissimilarity between all support samples. Therefore, in this paper we propose a negative-supervised capsule graph neural network (NSCGNN) which explicitly takes use of the similarity and dissimilarity between samples to make the text representations of the same type closer with each other and the ones of different types farther away, leading to representative and discriminative class prototypes. We firstly construct a graph to obtain text representations in the form of node capsules, where both intra-cluster similarity and inter-cluster dissimilarity between all samples are explored with information aggregation and negative supervision. Then, in order to induce generalized class prototypes based on those node capsules obtained from graph neural network, the dynamic routing algorithm is utilized in our model. Experimental results demonstrate the effectiveness of our proposed NSCGNN model, which outperforms existing few-shot approaches on three benchmark datasets.

scholarly journals Updating Awareness and Information Aggregation

2020 ◽  
Vol 0 (0) ◽  
Author(s):  
Spyros Galanis ◽  
Stelios Kotronis
Keyword(s):  
Mild Condition ◽  
Private Information ◽  
Dynamic Environment ◽  
Information Aggregation ◽  
Aggregate Information

AbstractThe ability of markets to aggregate information through prices is examined in a dynamic environment with unawareness. We find that if all traders are able to minimally update their awareness when they observe a price that is counterfactual to their private information, they will eventually reach an agreement, thus generalising the result of Geanakoplos and Polemarchakis (1982). Moreover, if the traded security is separable, then agreement is on the correct price and there is information aggregation, thus generalizing the result of Ostrovsky (2012) for non-strategic traders. We find that a trader increases her awareness if and only if she is able to become aware of something that other traders are already aware of and, under a mild condition, never becomes aware of anything more. In other words, agreement is more the result of understanding each other, rather than being unboundedly sophisticated.

Dynamic routing algorithm for avoiding hot spots in on-chip networks

2006 ◽  
Author(s):  
A. Sobhani ◽  
M. Daneshtalab ◽  
M.H. Neishaburi ◽  
M.D. Mottaghi ◽  
A. Afzali-Kusha ◽  
...  
Keyword(s):  
Hot Spots ◽  
Routing Algorithm ◽  
Dynamic Routing ◽  
On Chip

A Novel Energy-Balanced Dynamic Routing Algorithm for Wireless Sensor Networks

2014 ◽  
Vol 681 ◽  
pp. 235-238
Author(s):  
Xin Xin Zhou ◽  
Yan Zhao
Keyword(s):  
Wireless Sensor Networks ◽  
Sensor Networks ◽  
Data Transfer ◽  
Routing Algorithm ◽  
Dynamic Routing ◽  
Residual Energy ◽  
Wireless Sensor ◽  
The Novel ◽  
Balance Energy ◽  
Transfer Tasks

Wireless sensor networks (WSNs) is taking an increasing role in our lives. Because the energy of the sensors is limited how to efficiently use the energy to prolong the lifecycle of the sensor networks is very important. In this paper, a novel energy-balanced dynamic routing algorithm based on ACO is proposed. The novel routing algorithm can dynamically choose routing according to the residual energy of the sensors and the sensors with more power is taken more data transfer tasks. The simulation results show that the proposed routing algorithm can effectively balance energy consumption and prolong the lifecycle of the networks.

Sign in / Sign up

Export Citation Format

Share Document