I=
Pakistan Research Repository Home
 

Title of Thesis

Time Frequency Analysis Using Neural Networks

Author(s)

Imran Shafi

Institute/University/Department Details
Centre for Advance Studies in Engineering, Islamabad / University of Engineering and Technology, Taxila
Session
2009
Subject
Engineering Computer
Number of Pages
195
Keywords (Extracted from title, table of contents and abstract of thesis)
Neural, Diversity, Motivations, Analysis, Distributions, Exhibit, Frequency, Vectorization, Frequency, Time, Networks, Cluster, Experimental

Abstract
The thesis is divided in three parts. In the rst part, it explores and discusses the diversity of concepts and motivations for obtaining good resolution and highly concentrated timefrequency distributions (TFDs) for the research community.The description of the methods used for TFDs’ objective assessment is provided later in this part.
In the second part, a novel multiprocesses ANN based framework to obtain highly concentrated TFDs is proposed. The propose method utilizes a localised Bayesian regularised neural network model (BRNNM) to obtain the energy concentration along the instantaneous frequencies (IFs) of individual components in the multicomponent signals without assuming any prior knowledge. The spectrogram and preprocessed WignerVille distribution (WD) of the signals with known IF laws are used as the training set for the BRNNM. These distributions, taken as twodimensional (2D) image matrices, are vectorized and clustered according to the elbow criterion. Each cluster contains the pairs of the input and target vectors from the spectrograms and highly concentrated preprocessed WD respectively. For each cluster, the pairs of vectors are used to train the multiple ANNs under the Bayesian framework of David Mackay.The best trained network for each cluster is selected based on network error criterion. In the test phase, the test TFDs of unknown signals, after vectorization and clustering, are processed through these specialized ANNs. After postprocessing, the resulting TFDs are found to exhibit improved resolution and concentration along the individual components then the initial blurred estimates.
The third part presents the discussion on the experimental results obtained by the proposed technique. Moreover the framework is extended to include the various objective methods of assessment to evaluate the performance of deblurred TFDs obtained through the proposed technique. The selected methods not only allow quantifying the quality of TFDs instead of relying solely on visual inspection of their plots, but also help in drawing comparison of the proposed technique with the other existing techniques found in literature for the purpose. In particular the computation regularities show the effectiveness of the objective criteria in quantifying the TFDs’ concentration and resolution information.

Download Full Thesis
567 KB
S. No. Chapter Title of the Chapters Page Size (KB)
1 0 CONTENTS

 

 
82 KB
2

1

INTRODUCTION

1.1 Concentration and Resolution Tradeoff
1.2 Application speci c Environment
1.3 Objective Assessment of TFDs
1.4 Our Approach .
1.5 Contribution
1.6 Overview of the Thesis

1
196 KB
3 2 BACKGROUND

2.1 TF Analysis
2.2 Objective Assessment Methods
2.3 Summary

17
320 KB
4 3 NEURAL NETWORK BASED FRAMEWORK FOR COMPUTING DE.BLURRED TFDS.PART I

3.1 TFDs using ANN.Binary Case
3.2 Analysis & Comparison of the ANN Training Algorithms
3.3 Impact of varying number of Neurons and the Hidden Layers
3.4 Effect of Data Clustering and using Multiple ANNs for each Cluster
3.5 Summary

55
3,409 KB
5 4 NEURAL NETWORK BASED FRAMEWORK FOR COMPUTING DE.BLURRED TFDS.PART II

4.1 The ANN based Framework’s Description
4.2 Performance Evaluation
4.3 Summary

89
1,790 KB
6 5 DISCUSSION ON EXPERIMENTAL RESULTS

5.1 Visual Interpretation and Entropy Analysis
5.2 Objective Assessment
5.3 Summar

113
515 KB
7 6 CONCLUSION AND FUTURE DIRECTIONS

6.1 Future Directions

134
115 KB
8

7

REFERENCES

 

139
256 KB