论文句型参考

Ⅰ. 本文参考的论文

title :      Prediction Of Time Series Data Using GA-BPNN based Hybrid ANN Mode

author: Aishwarya D C , C.Narendra Babu

source: 2017 IEEE 7th International Advance Computing Conference

 

Ⅱ.key sentences stucture

 

2.1 Abstract

Key sentences:

1. Time series analysis has drawn the attention of many researchers as the findings from analysis are of huge importance and value.
2. Artificial Neural Networks are widely used for forecasting data because of its high capability to understand non linear relationships among data.
3. This paper aimed at hybridizing the traditional Back Propagation Neural Network (BPNN) with Genetic Algorithm(GA) to achieve better prediction accuracy.
4. The proposed model proves to produce better prediction results than the existing BPNN networks for both univariate and multivariate data sets.

Key sentence patterns：

1. …. has drawn the attention of many researchers as the findings from …. are of huge importance and value.
2. …. are widely used for … because of its high capability to …
3. This paper aimed at hybridizing … with … to achieve …
4. The proposed model proves to produce … than the existing …

 

2.2 Introduction

Key sentences:

1. Time Series Forecasting has always been one of the major research areas where new challenges arise and the scope of improvements of techniques is high.
2. Although there exists many models for time series prediction, there is no single unique model which can predict accurately,
3. One efficient nonlinear technique for time series forecasting is ANN.
4. If a single model is considered for such applications, the performance of the model and the prediction accuracy would not be as good as using a hybrid model.
5. In this paper, ANN based hybrid GA-BPNN model for data prediction is proposed and the model is evaluated and tested on four data sets consisting of both time series and multivariate data
6. The later part of the paper is organized as follows.
7. Section II discusses the existing models in the literature.
8. The paper is concluded in Section VI.
9. By strategically placing cloudlets in the WMAN and assigning user requests to cloudlets, we can minimize the total task request delay between users and cloudlets, bringing significant improvement to the performance of mobile applications
10. To make this concrete, consider the graph in Figure(5)

 

Key sentence patterns：

1. … has always been one of the major research areas where new challenge arise and the scope of improvements of techniques is high.
2. Although there exists many model for … , there is no …
3. One efficient … technique for … is …
4. If … is considered for such applications, the performance of the model and the … would be as good as using …
5. In this paper, … based … is proposed and model is evaluated and tested on …
6. The later part of the paper is organized as follows.
7. Section x discusses ….
8. The paper is concluded in Section …
9. By… we can …, bringing significant improvement to the performance of…
10. To make this concrete, consider the graph in …

 

 

 

2.3 related work

Key sentences:

1. In [1], the traditional ARIMA models were used for prediction of global temperature .
2. A hybrid of ARIMA model, ARIMA-GARCH was presented in [2].
3. The model was tested for forecasting Indian Stock and the hybrid model’s performance was better than the ARIMA model.
4. In [3] the traditional ARIMA model was compared with neural networks for commodity price forecasting and it was proved that neural networks perform better.
5. A hybrid ARIMA model based on ANN for forecasting different time series data is presented in [4].
6.  The different architectures of ANN, the feed forward network and the cascade network were considered in [5] for predicting the number of guests in a month.
7. A new approach called Multiplicative Neuron Model was used in [6] for non linear time series forecasting. For training this model Particle Swarm Optimization (PSO) was  made  use  of.
8. For the prediction of kWh output from a grid- connected photovoltaic system, ANN based PSO was used [19].
9. The concepts of Genetic Algorithm, when and where to use genetic algorithms are clearly described in [10].
10. The proposed model was compared with various other existing models and the proposed hybrid model proved to be more efficient.
11. Forecasting speed of wind was discussed in [13]  where wavelet based ANN was used and the model used criss cross optimization algorithm for training.
12. In [11] stock forecasting was done using fuzzy time series model  combined  with genetic algorithm.
13. Traffic flow was analyzed using different hybrid ANN models [20] like ARIMA model hybridized with ANN.
14. The same application in [19] was attempted using Bat algorithm based ANN model [21].
15. In order to obtain a minimal solution, Genetic Algorithm was made use of [22].

Key sentence patterns：

1. In  [参考文献],  the traditional … models based on … were used for …
2. A hybrid of … , … was presented in  [参考文献]
3. The model was tested for … and sth’s performance was better than …
4. In  [参考文献] the traditional … model was compared with … for … and it was proved that … performance better
5. … is presented in [参考文献]
6. … were considered in  [参考文献] for …
7. A new approach called … was used in  [参考文献] for …
8. For … , … based … was used in [参考文献]
9. …, when and where to use … are clearly described in [参考文献]
10. The proposed model was compared with various other existing models and the proposed … proved to be more efficient
11. … was discussed in [参考文献]  where … was used .
12. In [参考文献] … was done using … combined  with …
13. … was analyzed using …[参考文献]
14. The same application in [参考文献] was attempted using….
15. In order to … , …was made use of [参考文献].

 

 

2.4 Module and problem definition

    2.4.1 build a model

       … denote/imply/indicate/represent …    …表示…

Key sentences:

1. ANN is a non linear model which can be used for a wide range of applications.
2. The architecture resembles the human brain with interconnected neurons that process information.
3. Raw data which can be non linear is fed as input to the network, the hidden layer computes the weights of the input using an activation function or the transfer function.
4. LMA are used to solve non linear least squares problems.
5. Because of the use of second derivatives of the cost function, LM algorithm is robust than the GN algorithm which can converge faster and results in good solutions.
6. The algorithm is described using the following equations:
7. Genetic algorithms are inspired by biological computations.
8. It can also be defined as the algorithm which works on Darwin’s principle of “Survival of the fittest”.
9. The relationship between the binary strings consisting of zeros and ones is found using an evaluation function which returns a fitness value for that particular string.
10. There are various strategies for selecting, these include roulette wheel selection, tournament selection, selection based on ranking etc.
11. As discussed in Section I, the proposed model combines GA and BPNN models.
12. In the process of selection, the inputs with high fitness values are selected and they are used in further generations to get the best fit results.
13. The important parameters that has to be considered while using Genetic Algorithm are, Population  Size, Number of Generations, Crossover rate and Mutation rate.
14. in order to back propagate the network values, LM algorithm is made use of.
15. The BPNN network is designed with 2 hidden layers and the first hidden layer consists of 6 neurons where as the second hidden layer consists of a single neuron.
16. The maximum number of epochs is limited to  15000,  the momentum term is kept at 0.005 with a step increment of 10.
17. The training goes on until the minimum value of MSE is reached or when any of the other parameter conditions are met, say for example if the value of momentum maximum is reached, the training stops resulting in the best possible solution.

Key sentence patterns：

1. … is a … model which can be used for a wide range of applications.
2. The architecture resembles the ….
3. … is fed as input to the network, … computes … of the input using an … function or the … function.
4. … is used to solve … problems.
5. Because of …, … is robust than … which can converge faster and results in good solutions.
6. The algorithm is described using the following equations:
7. … algorithms are inspired by … computations
8. It can also be defined as the algorithm which works on Darwin’s principle of “Survival of the fittest”.
9. The relationship between …is found using an … function which returns….
10. There are various strategies for selecting, these include … and … etc
11. As discussed in Section I, the proposed model combines … and … models
12. In the process of …, the inputs with high fitness values are selected and they are used in further generations to get the best fit results.
13. The important parameters that has to be considered while using … Algorithm are, … and … (parameters)
14. in order to …, … algorithm is made use of
15. The BPNN network is designed with 2 hidden layers and the first hidden layer consists of 6 neurons where as the second hidden layer consists of a single neuron.
16. The maximum number of … is limited to 15000, … is kept at 0.005 with a step increment of 10
17. The training goes on until … is reached or when any of the other parameter conditions are met, say for example if … is reached, the training stops resulting in the best possible solution.

 

2.5 Results

Key sentences:

1. Four Data Sets were used for testing the performance of both BPNN and hybrid model.
2. Table.2 tabulates the performance results of all the data sets.
3. The results along with its performance measures are tabulated in Table.2
4. Figure.3 and 4 represents the raw data and the predicted results using BPNN and GA-BPNN models respectively.
5. The hybrid model performs better which is observed in Table.2.
6. From Table.2 it can be seen that the proposed model predicts better than the existing BPNN model.
7. As observed in the previous cases, the proposed model predicts better even for multivariate data. Results are tabulated in Table.2
8. From Table 2 it can be observed that the proposed hybrid model has better prediction results compared to BPNN networks.
9. The results are tabulated in Table 2 from where it can be observed that the proposed model performs better.
10. Prediction accuracy also depends on the size of the training set which is also evident form Table 1.

Key sentence patterns：

1. ... Data Sets were used for testing the performance of …
2. Table.2 tabulates the performance results of all the data sets
3. … along with … are tabulated in Table.2
4. Figure.3 and 4 represents … and … respectively
5. … performs better which is observed in Table.2
6. From Table.2 it can be seen that the proposed model … better than the existing … model.
7. As observed in … , the proposed model … better even for ….
8. From Table 2 it can be observed that the proposed … model has better prediction results compared to ….
9. The results are tabulated in Table 2 from where it can be observed that the proposed model performs better
10. … which is also evident form Table 1

 

2.6 Conclusion  

Key sentences:

1. Genetic Algorithm is very efficient in optimizing parameters, ANN is a widely accepted model for Time Series Prediction.
2. In all of the four cases, irrespective of univariate and multivariate data, the proposed model performed better.
3. The factors such as prediction horizon and the size of training set data were varied and the results were measured using MSE, MAPE and RMSE.
4. There is a significant improvement in the performance of the hybrid model.
5. Given the above four cases, the proposed model fits better than the BPNN model.
6. Although the proposed model has better performance, the time taken for convergence varies with the population size and the number of generations.
7. The random selection process that occurs during mutation is still not clear.
8. There is scope for further combinations.

Key sentence patterns：

1. … is very efficient in …, … is a widely accepted model for …
2. In all of the four cases, irrespective of … and …, the proposed model performed better
3. The factors such as …were varied and the results were measured using ….
4. There is a significant improvement in the performance of the … model.
5.  Given the above four cases, the proposed model fits better than ….
6. Although the proposed model has better performance, … taken for convergence varies with the population size and the number of generations.
7. … is still not clear.
8. There is scope for further combinations.