2021 International Conference on Circuits, Controls and Communications (CCUBE), 2021
Global Warming is one of the major climatic challenges. Land Surface Temperature (LST) is a manif... more Global Warming is one of the major climatic challenges. Land Surface Temperature (LST) is a manifestation of global warming and this happens due to the human activities on the earth. This work aims to correlate the LST with the Land Use Land Cover (LULC), vegetation density of the region and the elevation. The LULC and the vegetation index of the area have been derived using the optical remotely sensed images, whereas the LST is derived from the thermal images. The results of this research show that, for the given study area, the LST is not only dependent on the vegetation density and LULC, but it also has a complex relationship with the elevation of a geographic area. In other words, the LST is highest for the built-up areas, where the NDVI value is and it is least for the greener areas indicated by a higher value of the NDVI.
An image based approach which detects Indian coins of different denomination has been proposed in... more An image based approach which detects Indian coins of different denomination has been proposed in this paper. This consists of matching an input coin image with a database of coin images (templates) in two phases. The first phase involves, identifying matching radius database coins. In the second phase, template matching is performed by correlating the edges of input and matching radius database coin images. Template matching phase involves two parts, coarse matching and fine matching. This provides rotation invariance and does away with the requirement of placing the front face of the coin up. If the correlation coefficient obtained by template matching satisfies the threshold, then coin stands recognized. The algorithm has been developed in MATLAB 7.9.0 and the obtained results are recorded. The proposed method has been compared with existing methods (Difference, LBP, FFT) and comparison results have been recorded.
Serial communication is a technique where the transfer of data happens serially bitwise. Parallel... more Serial communication is a technique where the transfer of data happens serially bitwise. Parallel communication is a technique where the transfer of data happens parallelly simultaneously bitwise. In many applications serial communication is preferred. Comparison between I2C, SPI and CAN serial communication protocol can be seen. I2C communication protocol is preferred over SPI and CAN protocols. I2C is a synchronous serial communication which can be implemented in Verilog.
Humans are more likely to consider computers to be human-like when those computers understand and... more Humans are more likely to consider computers to be human-like when those computers understand and display appropriate nonverbal communicative behaviour. Our work presents an approach to automatic visual recognition of expressive face suitable for use in a vision-based affective framework. After describing the feature extraction techniques, classification results from several subjects are presented. Firstly, individual classifiers are trained separately with face features for classification into Face Action Units (FAU). Secondly, the same procedure is applied for classification into labelled emotion categories. A combined Ekman-Frison and Cohn-Kanade face action units are measured to label the emotion class. The five emotion classes like happy, sad, neutral, angry and surprise are considered in our analysis. The Open CV Library functions which are mainly aimed at real time computer vision are used extensively in our work. Depending on the various emotions detected, system responds to...
In any communication, security is the most important issue in today"s world. Lots of data se... more In any communication, security is the most important issue in today"s world. Lots of data security and data hiding algorithms have been developed in the last decade, which worked as motivation for the research. This project is a combination of steganography and cryptography, which provides a strong backbone for its security. The scenario of present day of information security system includes confidentiality, authenticity, integrity, non-repudiation. This present work focus is enlightening the technique to secure data or message with authenticity and integrity.In this project work, the secret message is encrypted before the actual embedding process starts. The entire work has done in MATLAB. The hidden message is encrypted using a simple encryption algorithm using secret key and hence it will be almost impossible for the intruder to unhide the actual secret message from the embedded cover file without knowing secret key.Only receiver and sender know the secret key.N-bit LSB subs...
International Journal of Engineering Trends and Technology, 2017
fMRI is a valuable experimental and diagnostic tool for assessing the human body especially the b... more fMRI is a valuable experimental and diagnostic tool for assessing the human body especially the brain. It has emerged as a successful tool in the investigation of cognitive functions. fMRI data was traditionally analysed with the univariate method, the popular one being Statistical Parametric Mapping based on the General Linear Model. But lately MVPA has been used to perform multivariate analysis of fMRI data. The multivariate approach originates from a field called as Machine Learning which is a branch of Artificial Intelligence. The Multivariate approaches have several advantages over the univariate approach, in that the Artificial Neural Networks (ANN) have outperformed some of the other classifiers such as Gaussian Naive Bayes, ICA and others. In this paper, an attempt is made to survey MVPA analysis of brain fMRI data using Artificial Neural Networks.
Handbook of Research on Synthesizing Human Emotion in Intelligent Systems and Robotics
It is essential to distinguish between an imposter and a genuine emotion in certain applications.... more It is essential to distinguish between an imposter and a genuine emotion in certain applications. To facilitate this, the number of features is increased by incorporating physiological signals. Physiological changes in the human body cannot be pretended. Human emotional behavior changes the heart rate, skin resistance, finger temperature, EEG etc. These physiological signal parameters can be measured and included as the final feature vector. The network is to be trained considering all the feature points as inputs with a radial basis activation function at the hidden layer and a linear activation function at the output layer. The two physiological parameters galvanic skin response (GSR) and finger tip temperature (FTT) that are predominant in deciding the emotion of a person are considered in this chapter. The measurements made are transmitted to LabVIEW add-on card for further data processing and analysis. The results obtained are nearer to the reality with a good measure of accuracy.
Advances in Intelligent Systems and Computing, 2013
An image based approach which detects Indian coins of different denomination has been proposed in... more An image based approach which detects Indian coins of different denomination has been proposed in this paper. This consists of matching an input coin image with a database of coin images (templates) in two phases. The first phase involves, identifying matching radius database coins. In the second phase, template matching is performed by correlating the edges of input and matching radius database coin images. Template matching phase involves two parts, coarse matching and fine matching. This provides rotation invariance and does away with the requirement of placing the front face of the coin up. If the correlation coefficient obtained by template matching satisfies the threshold, then coin stands recognized. The algorithm has been developed in MATLAB 7.9.0 and the obtained results are recorded. The proposed method has been compared with existing methods (Difference, LBP, FFT) and comparison results have been recorded.
International Journal of Innovative Technology and Exploring Engineering, 2019
The key to proper governance of the municipal bodies lies in knowing the geography of the region.... more The key to proper governance of the municipal bodies lies in knowing the geography of the region. The land cover of the region changes with respect to time. Also, there are seasonal variation in the layout of the waterbodies. Manual verification and surveying of these things becomes very difficult for want of resources. Remote Sensing Images play a very important role in mapping the land cover. In this paper, we consider such remotely sensed Multispectral Images, taken from Landsat-8. Parametric Machine learning algorithm like Maximum Likelihood Classifier has been used on those images to classify the land cover. Normalized Difference Vegetation Index (NDVI) has been calculated and integrates with the classification process. Four basic land covers have been identified for the purpose namely Water, Vegetation, Built-up and Barren soil. The area of study is Bangalore urban region where we find that the water bodies are decreasing day by day. An overall efficiency of 82% with a kappa h...
Journal of the Indian Society of Remote Sensing, 2021
Urbanization plays a major role in the governance policies of the civic bodies. Predicting the gr... more Urbanization plays a major role in the governance policies of the civic bodies. Predicting the growth of the cities helps the government to provide the infrastructure facilities. Getting to know the number of built-up areas indirectly points to population increase or decrease. The percentage of the land occupied by built-up areas or non-built-up areas can be found out using the remotely sensed images. Land use and land cover (LULC) of these images point out how the land has been used, whether the land is a water body, vegetation, soil or built-up area. This work explores the combination of the remote sensing images and the use multilayer perceptron (MLP)-based cellular automata (CA) for predicting the changes in the LULC on a later date, using two different initial LULC maps as the input. Four basic classes, namely water, vegetation, built-up and soil, have been considered in this work. Machine learning algorithm (Random Forest Classifier) is used to obtain the LULC map of all the 3 years under consideration, namely 2013, 2016 and 2019. The LULC map of the years 2013 and 2016 acts as the initial rasters for the prediction, whereas the LULC map of the year 2019 is used for validation. The novelty of the work lies in the use of Normalized Difference Built-up index raster along with the roadway map, as the spatial variables in the MLP-based CA. The predicted results show an increase of 12.65% in the built-up areas, decrease of the soil class by 5.23%, decrease of the vegetation by 6.63% and decrease in the water bodies by 0.79% compared the base year considered (2013). The predicted LULC map using the proposed model is then validated against an independently obtained LULC map of 2019. The Jaccard similarity coefficient was found to be 0.93, which indicates the validity of the algorithm for future predictions.
2009 2nd International Congress on Image and Signal Processing, 2009
In online applications, authentication systems which are covenant with a measurable behavioral tr... more In online applications, authentication systems which are covenant with a measurable behavioral trait and physiological characteristics are essential. This paper deals with authentication of an individual's on-line signature data and textual iris information using continuous dynamic programming [CDP]. Instead of working on primary features such as image or on-line data, working on the derived kinematic plot is robust way of
In a multi script environment, an archive of documents printed in different scripts is in practic... more In a multi script environment, an archive of documents printed in different scripts is in practice. For automatic processing of such documents through Optical Character Recognition (OCR), it is necessary to identify the script type of the document. In this paper, a novel texture-based approach is presented to identify the script type of the collection of documents printed in ten Indian scripts Bangla, Devanagari, Roman (English), Gujarati, Malayalam, Oriya, Tamil, Telugu, Kannada and Urdu. The document images are decomposed through the Wavelet Packet Decomposition using the Haar basis function up to level two. Gray level co-occurrence matrix is constructed for the coefficient sub bands of the wavelet transform. The Haralick texture features are extracted from the cooccurrence matrix and then used in the identification of the script of a machine printed document. Experimentation conducted involved 3000 text images for learning and 2500 text images for testing. Script classification p...
The need for security has been increasing day by day. Various algorithms for encryption and decry... more The need for security has been increasing day by day. Various algorithms for encryption and decryption have been proposed, but all are having some problem with them. Some having less security, some consumes more area, some are having less speed and throughput. So our objective is to propose high effective AES core hardware architecture for implementing it to encrypt/decrypt the data in portable hard disk drive system that apply to effectively in the terms of speed, scale size and power consumption to comply with minimum speed of 5 Gbps (USB3.0). The 128 bits data path of two different AES architectures design has been proposed. Basic Iterative AES reuses the same hardware for all the ten iterations. One Stage Sub Pipelined AES, with one stage of outer pipelining in the data blocks that both of them are purely 128 bits data path architecture that different from the previous public paper.
Objective imaging-based biomarker discovery for psychiatric conditions is critical for accurate d... more Objective imaging-based biomarker discovery for psychiatric conditions is critical for accurate diagnosis and treatment. Using a machine learning framework, this work investigated the utility of brain's functional network topology (complex network features) extracted from functional magnetic resonance imaging (fMRI) functional connectivity (FC) as viable biomarker of autism spectrum disorder (ASD). To this end, we utilized resting-state fMRI data from the publicly available ABIDE dataset consisting of 432 ASD patients and 556 matched healthy controls. Upon standard pre-processing, 3D + time fMRI data were parcellated into 200 functionally homogenous regions, and whole-brain FC network using Pearson's correlation was obtained from corresponding regional mean time series. A battery of complex network features were computed from the FC network using graph theoretic techniques. Recursive-Cluster-Elimination Support Vector Machine algorithm was employed to compare the predictive performance of three independent feature sets, (i) FC, (ii) complex network measures, and (iii) both combined. The study found that FC could diagnose ASD with 67.3 % accuracy and graph measures with 64.5 % accuracy, while the combined feature set could diagnose with 70.1 % accuracy (all accuracies were significantly different, p < 10 − 30). The most discriminative imaging features were mainly from lateral temporal, occipital, precuneus (all reduced in ASD) and orbito-frontal (elevated in ASD) regions. We concluded that network topology (graph measures) carried some unique information about ASD pathology not available in bivariate connectivity (FC), and that using both together provided better prediction than using individual measures. Future prediction studies could incorporate multiple fMRI analysis strategies within their framework to achieve superior prediction performances.
Hybrid hierarchical clustering techniques which combine the characteristics of different partitio... more Hybrid hierarchical clustering techniques which combine the characteristics of different partitional clustering techniques or partitional and hierarchical clustering techniques are interesting. In this paper, efficient bottom-up hybrid hierarchical clustering (BHHC) techniques have been proposed for the purpose of prototype selection for protein sequence classification. In the first stage, an incremental partitional clustering technique such as leader algorithm (ordered leader no update (OLNU) method) which requires only one database (db) scan is used to find a set of subcluster representatives. In the second stage, either a hierarchical agglomerative clustering (HAC) scheme or a partitional clustering algorithm-'K-medians' is used on these subcluster representatives to obtain a required number of clusters. Thus, this hybrid scheme is scalable and hence would be suitable for clustering large data sets and we also get a hierarchical structure consisting of clusters and subclusters and the representatives of which are used for pattern classification. Even if more number of prototypes are generated, classification time does not increase much as only a part of the hierarchical structure is searched. The experimental results (classification accuracy (CA) using the prototypes obtained and the computation time) of the proposed algorithms are compared with that of the hierarchical agglomerative schemes, K-medians and nearest neighbour classifier (NNC) methods. The proposed methods are found to be computationally efficient with reasonably good CA.
2021 International Conference on Circuits, Controls and Communications (CCUBE), 2021
Global Warming is one of the major climatic challenges. Land Surface Temperature (LST) is a manif... more Global Warming is one of the major climatic challenges. Land Surface Temperature (LST) is a manifestation of global warming and this happens due to the human activities on the earth. This work aims to correlate the LST with the Land Use Land Cover (LULC), vegetation density of the region and the elevation. The LULC and the vegetation index of the area have been derived using the optical remotely sensed images, whereas the LST is derived from the thermal images. The results of this research show that, for the given study area, the LST is not only dependent on the vegetation density and LULC, but it also has a complex relationship with the elevation of a geographic area. In other words, the LST is highest for the built-up areas, where the NDVI value is and it is least for the greener areas indicated by a higher value of the NDVI.
An image based approach which detects Indian coins of different denomination has been proposed in... more An image based approach which detects Indian coins of different denomination has been proposed in this paper. This consists of matching an input coin image with a database of coin images (templates) in two phases. The first phase involves, identifying matching radius database coins. In the second phase, template matching is performed by correlating the edges of input and matching radius database coin images. Template matching phase involves two parts, coarse matching and fine matching. This provides rotation invariance and does away with the requirement of placing the front face of the coin up. If the correlation coefficient obtained by template matching satisfies the threshold, then coin stands recognized. The algorithm has been developed in MATLAB 7.9.0 and the obtained results are recorded. The proposed method has been compared with existing methods (Difference, LBP, FFT) and comparison results have been recorded.
Serial communication is a technique where the transfer of data happens serially bitwise. Parallel... more Serial communication is a technique where the transfer of data happens serially bitwise. Parallel communication is a technique where the transfer of data happens parallelly simultaneously bitwise. In many applications serial communication is preferred. Comparison between I2C, SPI and CAN serial communication protocol can be seen. I2C communication protocol is preferred over SPI and CAN protocols. I2C is a synchronous serial communication which can be implemented in Verilog.
Humans are more likely to consider computers to be human-like when those computers understand and... more Humans are more likely to consider computers to be human-like when those computers understand and display appropriate nonverbal communicative behaviour. Our work presents an approach to automatic visual recognition of expressive face suitable for use in a vision-based affective framework. After describing the feature extraction techniques, classification results from several subjects are presented. Firstly, individual classifiers are trained separately with face features for classification into Face Action Units (FAU). Secondly, the same procedure is applied for classification into labelled emotion categories. A combined Ekman-Frison and Cohn-Kanade face action units are measured to label the emotion class. The five emotion classes like happy, sad, neutral, angry and surprise are considered in our analysis. The Open CV Library functions which are mainly aimed at real time computer vision are used extensively in our work. Depending on the various emotions detected, system responds to...
In any communication, security is the most important issue in today"s world. Lots of data se... more In any communication, security is the most important issue in today"s world. Lots of data security and data hiding algorithms have been developed in the last decade, which worked as motivation for the research. This project is a combination of steganography and cryptography, which provides a strong backbone for its security. The scenario of present day of information security system includes confidentiality, authenticity, integrity, non-repudiation. This present work focus is enlightening the technique to secure data or message with authenticity and integrity.In this project work, the secret message is encrypted before the actual embedding process starts. The entire work has done in MATLAB. The hidden message is encrypted using a simple encryption algorithm using secret key and hence it will be almost impossible for the intruder to unhide the actual secret message from the embedded cover file without knowing secret key.Only receiver and sender know the secret key.N-bit LSB subs...
International Journal of Engineering Trends and Technology, 2017
fMRI is a valuable experimental and diagnostic tool for assessing the human body especially the b... more fMRI is a valuable experimental and diagnostic tool for assessing the human body especially the brain. It has emerged as a successful tool in the investigation of cognitive functions. fMRI data was traditionally analysed with the univariate method, the popular one being Statistical Parametric Mapping based on the General Linear Model. But lately MVPA has been used to perform multivariate analysis of fMRI data. The multivariate approach originates from a field called as Machine Learning which is a branch of Artificial Intelligence. The Multivariate approaches have several advantages over the univariate approach, in that the Artificial Neural Networks (ANN) have outperformed some of the other classifiers such as Gaussian Naive Bayes, ICA and others. In this paper, an attempt is made to survey MVPA analysis of brain fMRI data using Artificial Neural Networks.
Handbook of Research on Synthesizing Human Emotion in Intelligent Systems and Robotics
It is essential to distinguish between an imposter and a genuine emotion in certain applications.... more It is essential to distinguish between an imposter and a genuine emotion in certain applications. To facilitate this, the number of features is increased by incorporating physiological signals. Physiological changes in the human body cannot be pretended. Human emotional behavior changes the heart rate, skin resistance, finger temperature, EEG etc. These physiological signal parameters can be measured and included as the final feature vector. The network is to be trained considering all the feature points as inputs with a radial basis activation function at the hidden layer and a linear activation function at the output layer. The two physiological parameters galvanic skin response (GSR) and finger tip temperature (FTT) that are predominant in deciding the emotion of a person are considered in this chapter. The measurements made are transmitted to LabVIEW add-on card for further data processing and analysis. The results obtained are nearer to the reality with a good measure of accuracy.
Advances in Intelligent Systems and Computing, 2013
An image based approach which detects Indian coins of different denomination has been proposed in... more An image based approach which detects Indian coins of different denomination has been proposed in this paper. This consists of matching an input coin image with a database of coin images (templates) in two phases. The first phase involves, identifying matching radius database coins. In the second phase, template matching is performed by correlating the edges of input and matching radius database coin images. Template matching phase involves two parts, coarse matching and fine matching. This provides rotation invariance and does away with the requirement of placing the front face of the coin up. If the correlation coefficient obtained by template matching satisfies the threshold, then coin stands recognized. The algorithm has been developed in MATLAB 7.9.0 and the obtained results are recorded. The proposed method has been compared with existing methods (Difference, LBP, FFT) and comparison results have been recorded.
International Journal of Innovative Technology and Exploring Engineering, 2019
The key to proper governance of the municipal bodies lies in knowing the geography of the region.... more The key to proper governance of the municipal bodies lies in knowing the geography of the region. The land cover of the region changes with respect to time. Also, there are seasonal variation in the layout of the waterbodies. Manual verification and surveying of these things becomes very difficult for want of resources. Remote Sensing Images play a very important role in mapping the land cover. In this paper, we consider such remotely sensed Multispectral Images, taken from Landsat-8. Parametric Machine learning algorithm like Maximum Likelihood Classifier has been used on those images to classify the land cover. Normalized Difference Vegetation Index (NDVI) has been calculated and integrates with the classification process. Four basic land covers have been identified for the purpose namely Water, Vegetation, Built-up and Barren soil. The area of study is Bangalore urban region where we find that the water bodies are decreasing day by day. An overall efficiency of 82% with a kappa h...
Journal of the Indian Society of Remote Sensing, 2021
Urbanization plays a major role in the governance policies of the civic bodies. Predicting the gr... more Urbanization plays a major role in the governance policies of the civic bodies. Predicting the growth of the cities helps the government to provide the infrastructure facilities. Getting to know the number of built-up areas indirectly points to population increase or decrease. The percentage of the land occupied by built-up areas or non-built-up areas can be found out using the remotely sensed images. Land use and land cover (LULC) of these images point out how the land has been used, whether the land is a water body, vegetation, soil or built-up area. This work explores the combination of the remote sensing images and the use multilayer perceptron (MLP)-based cellular automata (CA) for predicting the changes in the LULC on a later date, using two different initial LULC maps as the input. Four basic classes, namely water, vegetation, built-up and soil, have been considered in this work. Machine learning algorithm (Random Forest Classifier) is used to obtain the LULC map of all the 3 years under consideration, namely 2013, 2016 and 2019. The LULC map of the years 2013 and 2016 acts as the initial rasters for the prediction, whereas the LULC map of the year 2019 is used for validation. The novelty of the work lies in the use of Normalized Difference Built-up index raster along with the roadway map, as the spatial variables in the MLP-based CA. The predicted results show an increase of 12.65% in the built-up areas, decrease of the soil class by 5.23%, decrease of the vegetation by 6.63% and decrease in the water bodies by 0.79% compared the base year considered (2013). The predicted LULC map using the proposed model is then validated against an independently obtained LULC map of 2019. The Jaccard similarity coefficient was found to be 0.93, which indicates the validity of the algorithm for future predictions.
2009 2nd International Congress on Image and Signal Processing, 2009
In online applications, authentication systems which are covenant with a measurable behavioral tr... more In online applications, authentication systems which are covenant with a measurable behavioral trait and physiological characteristics are essential. This paper deals with authentication of an individual&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#x27;s on-line signature data and textual iris information using continuous dynamic programming [CDP]. Instead of working on primary features such as image or on-line data, working on the derived kinematic plot is robust way of
In a multi script environment, an archive of documents printed in different scripts is in practic... more In a multi script environment, an archive of documents printed in different scripts is in practice. For automatic processing of such documents through Optical Character Recognition (OCR), it is necessary to identify the script type of the document. In this paper, a novel texture-based approach is presented to identify the script type of the collection of documents printed in ten Indian scripts Bangla, Devanagari, Roman (English), Gujarati, Malayalam, Oriya, Tamil, Telugu, Kannada and Urdu. The document images are decomposed through the Wavelet Packet Decomposition using the Haar basis function up to level two. Gray level co-occurrence matrix is constructed for the coefficient sub bands of the wavelet transform. The Haralick texture features are extracted from the cooccurrence matrix and then used in the identification of the script of a machine printed document. Experimentation conducted involved 3000 text images for learning and 2500 text images for testing. Script classification p...
The need for security has been increasing day by day. Various algorithms for encryption and decry... more The need for security has been increasing day by day. Various algorithms for encryption and decryption have been proposed, but all are having some problem with them. Some having less security, some consumes more area, some are having less speed and throughput. So our objective is to propose high effective AES core hardware architecture for implementing it to encrypt/decrypt the data in portable hard disk drive system that apply to effectively in the terms of speed, scale size and power consumption to comply with minimum speed of 5 Gbps (USB3.0). The 128 bits data path of two different AES architectures design has been proposed. Basic Iterative AES reuses the same hardware for all the ten iterations. One Stage Sub Pipelined AES, with one stage of outer pipelining in the data blocks that both of them are purely 128 bits data path architecture that different from the previous public paper.
Objective imaging-based biomarker discovery for psychiatric conditions is critical for accurate d... more Objective imaging-based biomarker discovery for psychiatric conditions is critical for accurate diagnosis and treatment. Using a machine learning framework, this work investigated the utility of brain's functional network topology (complex network features) extracted from functional magnetic resonance imaging (fMRI) functional connectivity (FC) as viable biomarker of autism spectrum disorder (ASD). To this end, we utilized resting-state fMRI data from the publicly available ABIDE dataset consisting of 432 ASD patients and 556 matched healthy controls. Upon standard pre-processing, 3D + time fMRI data were parcellated into 200 functionally homogenous regions, and whole-brain FC network using Pearson's correlation was obtained from corresponding regional mean time series. A battery of complex network features were computed from the FC network using graph theoretic techniques. Recursive-Cluster-Elimination Support Vector Machine algorithm was employed to compare the predictive performance of three independent feature sets, (i) FC, (ii) complex network measures, and (iii) both combined. The study found that FC could diagnose ASD with 67.3 % accuracy and graph measures with 64.5 % accuracy, while the combined feature set could diagnose with 70.1 % accuracy (all accuracies were significantly different, p < 10 − 30). The most discriminative imaging features were mainly from lateral temporal, occipital, precuneus (all reduced in ASD) and orbito-frontal (elevated in ASD) regions. We concluded that network topology (graph measures) carried some unique information about ASD pathology not available in bivariate connectivity (FC), and that using both together provided better prediction than using individual measures. Future prediction studies could incorporate multiple fMRI analysis strategies within their framework to achieve superior prediction performances.
Hybrid hierarchical clustering techniques which combine the characteristics of different partitio... more Hybrid hierarchical clustering techniques which combine the characteristics of different partitional clustering techniques or partitional and hierarchical clustering techniques are interesting. In this paper, efficient bottom-up hybrid hierarchical clustering (BHHC) techniques have been proposed for the purpose of prototype selection for protein sequence classification. In the first stage, an incremental partitional clustering technique such as leader algorithm (ordered leader no update (OLNU) method) which requires only one database (db) scan is used to find a set of subcluster representatives. In the second stage, either a hierarchical agglomerative clustering (HAC) scheme or a partitional clustering algorithm-'K-medians' is used on these subcluster representatives to obtain a required number of clusters. Thus, this hybrid scheme is scalable and hence would be suitable for clustering large data sets and we also get a hierarchical structure consisting of clusters and subclusters and the representatives of which are used for pattern classification. Even if more number of prototypes are generated, classification time does not increase much as only a part of the hierarchical structure is searched. The experimental results (classification accuracy (CA) using the prototypes obtained and the computation time) of the proposed algorithms are compared with that of the hierarchical agglomerative schemes, K-medians and nearest neighbour classifier (NNC) methods. The proposed methods are found to be computationally efficient with reasonably good CA.
Uploads
Papers by P. A. Vijaya