Scribd
Upload
33 views

Time Series Prediction Based On Ensemble ANFIS

Uploaded by

Sushma V
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
33 views

Time Series Prediction Based On Ensemble ANFIS

Uploaded by

Sushma V
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

Introduction to Scientific Writing

Time series prediction based on ensemble ANFIS


1. REFERENCE
De-Wang Chen and Jun-Ping Zhang, "Time series prediction based on ensemble ANFIS,"
2005 International Conference on Machine Learning and Cybernetics, Guangzhou, China,
2005, pp. 3552-3556 Vol. 6, doi: 10.1109/ICMLC.2005.1527557.

2. ABSTRACT
In this paper, random and bootstrap sampling method and ANFIS (adaptive network based
fuzzy inference system) are integrated into En-ANFIS (an ensemble ANFIS) to predict chaotic
and traffic flow time series. The prediction results of En-ANFIS are compared with an ANFIS
using all training data and each ANFIS unit in En-ANFIS. Experimental results show that the
prediction accuracy of the En-ANFIS is higher than that of single ANFIS unit, while the
number of training sample and training time of the En-ANFIS are less than that of the ANFIS
using all training data. So, En-ANFIS is an effective method to achieve both high accuracy and
less computational complexity for time series prediction.

Keywords: Time series prediction; ANFIS; ensemble learning; bootstrap; traffic flow

3. SUMMARY
 Time series prediction is a branch of statistical and probability discipline with many
applications in traffic flow prediction, weather analysis, and economics prediction.
 There are many approaches for time series prediction:
 Kalman filtering, neural network, linear regression, and fuzzy system.
 Linear regression is a simple approach but is less adaptive.
 Kalman filtering is an adaptive approach but linear.
 The neural network approximates nonlinear function, but it requires plenty of data
training and is challenging to interpret.
 On the contrary, the fuzzy system has excellent interpretation capability but has
relatively low adaptability.
 ANFIS has the integrated advantage of both fuzzy systems and neural networks,
which has excellent learning capability and also easily interpreted.
 Applications of ANFIS are time series prediction, intelligent control, and function
approximation.
 In this paper author integrates the ensemble learning and ANFIS to En-ANFIS time
series prediction algorithm and comparison of the results of the time series
prediction using ANFIS and En-ANFIS is made.
 ANFIS, the neural network implementation of the fuzzy system, is the hybrid
algorithm, which is the combination of Backpropagation and least square estimation
algorithm to have a great deal of learning speed.
 Ensemble is a learning model where multiple learners are trained for an identical
task, and the predictions of these learners are merged for dealing with later
instances.
 En-ANFIS is segmented into five layers: an input layer, sample layer, training layer,
testing layer, and output layer.
 In the training layer, each component of ANFIS is trained using randomly selected or
bootstrapped training data.
 In the testing layer, the output of each trained component of ANFIS is tested
separately with the testing data.
 The final output of En-ANFIS is the overall output of all ANFIS components in the
testing layer.
 In this paper, the output of En-ANFIS is calculated using two methods: uniform
weighting and non-uniform weighting.
 Though En-ANFIS is more complex than single ANFIS, since the computation happens
parallelly, the computation complexity does not increase.
 Approximation of the intrinsic distribution of data is guaranteed by multi-
bootstrapping or random sample.
 To validate the impact of the proposed En-ANFIS traffic flow time series or a chaotic
time series is used.
 The author conducts an experiment taking the Mackey-Glass time-series data and
traffic flow data to compare and analyze different algorithms.
 The tabulated results show that En-ANFIS is always better than ANFIS for any
sampling technology or weighting methods used.
 The prediction accuracy of En-ANFIS is the same as that of allANFIS. However, En-
ANFIS uses considerably less time and training data.
 The weighting methods and sampling methods have a trivial influence on the
prediction accuracy of En-ANFIS.
 This shows that when multiple weak ANFIS are ensembled high performance or
accuracy can be expected.
 En-ANFIS uses less training data and less training time to obtain great performance.
 Besides, the author claims that this method is not only efficient in the prediction
problem but also works well with pattern recognition, classification, and many other
domains.
4. NOTES
 Author gives a clear ground explanation of time series prediction algorithms that will
be discussed.
 Author makes a clear statement with the procedures used and why it is used.
 Learnt about ensemble learning and ANFIS. Pros and Cons of using these methods.
 How to overcome the cons by combing ensemble learning and ANFIS to achieve a
greater accuracy.
 Author gives a nice tabulation to shows the results of the experimentations, gives a
clarity of the experiment’s outcomes and makes it easier for the comparison.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy