Dr. Shahu Chatrapati - Academia.edu (original) (raw)

Papers by Dr. Shahu Chatrapati

Research paper thumbnail of Variable initial energy and unequal clustering (VEUC) based multicasting in WSN

2017 International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET), 2017

Multicast Communication plays an important role in most of the resource constrained networking en... more Multicast Communication plays an important role in most of the resource constrained networking environments such as Wireless Sensor Networks (WSN), Internet of Things (IOT). Communication in WSN is restricted by energy, computation and memory capabilities of sensor nodes. Designing an efficient routing algorithm to achieve communication between Stationary Base station (BS) and a cluster of sensor nodes in a specific region requires the base station to send individual messages to all sensor nodes. This approach consumes a large amount of energy and bandwidth. A variety of algorithms exist to address this issue by dividing the sensor nodes into clusters. Each cluster is monitored by a Cluster Head (CH), responsible for gathering and aggregating data to send the same to the BS. In this paper, we reviewed existing clustering techniques and propose an unequal clustering based scheme. This allows the BS to communicate a multicast message to cluster members as well as a cluster head to communicate with other cluster members. The results show that our approach improves network lifetime.

Research paper thumbnail of Energy-Efficient Neighbor Discovery Using Bacterial Foraging Optimization (BFO) Algorithm for Directional Wireless Sensor Networks

Lecture Notes in Electrical Engineering, 2021

In directional wireless sensor networks (WSN), the existing neighbor's discovery methods involve ... more In directional wireless sensor networks (WSN), the existing neighbor's discovery methods involve high latency and energy consumption, compared to the block design-based methods. Moreover, the duty cycle schedule of nodes has to be addressed to increase the network lifetime. In this paper, an energy-efficient collaborative neighbor discovery mechanism using the bacterial foraging optimization (BFO) algorithm is recommended. In this computation, each node with a directional antenna performs beamforming using BFOA with sector number and beam direction as the fitness function. In the end, appropriate active nodes with higher energy levels are selected from the neighbors during data transmission. The obtained results have shown that the recommended model minimizes power conservation and delay and enhances the lifetime of time of network activity.

Research paper thumbnail of Neighbor Nodes Discovery Schemes in a Wireless Sensor Network

Electronics and Communications Engineering, 2019

Research paper thumbnail of DEGSA-VMM: Dragonfly-based exponential gravitational search algorithm to VMM strategy for load balancing in cloud computing

Kybernetes, 2018

Purpose This paper aims to develop the Dragonfly-based exponential gravitational search algorithm... more Purpose This paper aims to develop the Dragonfly-based exponential gravitational search algorithm to VMM strategy for effective load balancing in cloud computing. Due to widespread growth of cloud users, load balancing is the essential criterion to deal with the overload and underload problems of the physical servers. DEGSA-VMM is introduced, which calculates the optimized position to perform the virtual machine migration (VMM). Design/methodology/approach This paper presents an algorithm Dragonfly-based exponential gravitational search algorithm (DEGSA) that is based on the VMM strategy to migrate the virtual machines of the overloaded physical machine to the other physical machine keeping in mind the energy, migration cost, load and quality of service (QoS) constraints. For effective migration, a fitness function is provided, which selects the best fit that possess minimum energy, cost, load and maximum QoS contributing toward the maximum energy utilization. Findings For the perfo...

Research paper thumbnail of ECC-Based Secure Group Communication in Energy-Efficient Unequal Clustered WSN (EEUC-ECC)

With an advent of the Internet of things (IoT), wireless sensor networks (WSNs) are gaining popul... more With an advent of the Internet of things (IoT), wireless sensor networks (WSNs) are gaining popularity in application areas like smart cities, body area sensor networks, industrial process control, and habitat and environment monitoring. Since these networks are exposed to various attacks like node compromise attack, DoS attacks, etc., the need for secured communication is evident. We present an updated survey on various secure group communication (SGC) schemes and evaluate their performance in terms of space and computational complexity. We also propose a novel technique for secure and scalable group communication that performs better compared with existing approaches.

Research paper thumbnail of Competitive Equilibrium Approach for Load Balancing a Computational Grid with Communication Delays

Computational grids interconnect hundreds of heterogeneous computing resources from geographicall... more Computational grids interconnect hundreds of heterogeneous computing resources from geographically remote sites, designed to meet the large demands of many users from scientific and business domains. A job initiated at one site can be executed by any of the computing resources. Therefore, response time of a job includes processing delay at the site of execution and communication delay for transferring the job from the site of initiation to the site of execution. Load balancing is allocation of jobs to available resources so as to optimize a given objective function. The objective can be achieving a system optimal solution, which tries to minimize the mean response time of all users or an individual optimal solution which tries to minimize each user’s response time. Previous works on load balancing either considered only system optimal objective or individual optimal objective. This paper introduces competitive equilibrium solution, a pricing mechanism for load balancing that indepen...

Research paper thumbnail of Competitive equilibrium approach for load balanicing a grid network

Research paper thumbnail of An Energy Efficient Node Scheduling based Congestion Control Scheme for WSN Multicasting

EAI Endorsed Transactions on Energy Web, 2018

Research paper thumbnail of A Study on Text Similarity Measures

Journal of Advanced Research in Dynamical and Control Systems, 2020

Research paper thumbnail of Competitive equilibrium based personal data market model

Indian Journal of Science and Technology, 2020

Background/Objectives: In a digital economy, with the increasing commercial value, personal data ... more Background/Objectives: In a digital economy, with the increasing commercial value, personal data is viewed as a commodity to be bought and sold. Data owners expect an appropriate compensation for trading off their privacy depending on how they value their privacy. Simultaneously, data consumers want to maximize their utility which is dependent on the value derived from the data. Consequently, a data market model that optimally recompenses data owners and maximizes the profits of data consumers is required. Methods: In this study, a data market model and pricing mechanism based on Fisher market model and competitive equilibrium are presented where the value derived from the data is calculated from information entropy. The proposed data model and the pricing mechanism jointly and simultaneously maximize the profit of data owners and the utility of data buyers. Findings: Experiments are conducted on adult data set to validate the efficacy of the proposed approach. Data owners are classified as risk averse, risk neutral, risk taking and privacy regarders. Subsequently, prices of data samples are adjusted to reach equilibrium as defined by the Fisher market model maximizing simultaneously and jointly the profit of data owners and the utility of data buyers. Applications: The proposed competitive equilibrium based personal data market model can be used to find equilibrium prices and bundles of data samples for each data buyer at these prices maximizing the utility of data buyer subject to his budgetary constraints and data requirements, and data owners' privacy preferences.

Research paper thumbnail of Exponential gravitational search algorithm-based VM migration strategy for load balancing in cloud computing

International Journal of Modeling, Simulation, and Scientific Computing, 2017

With the advancement in the science and technology, cloud computing has become a recent trend in ... more With the advancement in the science and technology, cloud computing has become a recent trend in environment with immense requirement of infrastructure and resources. Load balancing of cloud computing environments is an important matter of concern. The migration of the overloaded virtual machines (VMs) to the underloaded VM with optimized resource utilization is the effective way of the load balancing. In this paper, a new VM migration algorithm for the load balancing in the cloud is proposed. The migration algorithm proposed (EGSA-VMM) is based on exponential gravitational search algorithm which is the integration of gravitational search algorithm and exponential weighted moving average theory. In our approach, the migration is done based on the migration cost and QoS. The experimentation of proposed EGSA-based VM migration algorithm is compared with ACO and GSA. The simulation of experiments shows that the proposed EGSA-VMM algorithm achieves load balancing and reasonable resource...

Research paper thumbnail of New secure routing protocol with elliptic curve cryptography for military heterogeneous wireless sensor networks

Journal of Information and Optimization Sciences, 2017

Military Heterogeneous Wireless Sensor Network (MHTWSN) is an emerging technology. Due to its ext... more Military Heterogeneous Wireless Sensor Network (MHTWSN) is an emerging technology. Due to its extending services, MHTWSN is becoming a widely preferable network for border area monitoring and enemy object tracking in sensitive, unmanned areas. However, security is one of the key challenging issue in MHTWSN. Many routing protocols like AODV, AOMDV, I2MR, and IMCC have proposed for efficient routing establishment and data delivery, but these routing protocols results the poor performance when the network is under any security attack. So we have proposed a new secure multipath routing protocol

Research paper thumbnail of Multipath Interference Minimization in Heterogeneous Wireless Sensor Networks for Reliable Data Transfer

2016 International Conference on Computer and Communication Engineering (ICCCE), 2016

Multipath interference or route coupling is a severe issue in multipath routing schemes of wirele... more Multipath interference or route coupling is a severe issue in multipath routing schemes of wireless sensor networks. Due to that, quality of data transmission is impossible in wireless sensor network (WSN) or heterogeneous wireless sensor network (HTWSN). Multipath routing is having many features over single path routing, but it is highly prone to multipath interference. The existing multipath routing schemes can results poor network performance and low reliable data transmission due to the existence of multipath interference. So we propose interference minimization clustering multipath routing protocol (IMCMRP) for multipath interference minimization with energy efficiency and reliable data transmission. Simulation results shows that, IMCMRP protocol is achieved higher data delivery ratio 20% and 31%, better network life time 20 % and 42 %, lower latency up to 30 % and 60 %, higher throughput 30 % and 48 % than LIEMRO multipath routing and single path routing schemes.

Research paper thumbnail of Interference Minimization Protocol in Heterogeneous Wireless Sensor Networks for Military Applications

Smart Innovation, Systems and Technologies, 2016

Wireless sensor Network (WSN) is an emerging technology has significant applications in several i... more Wireless sensor Network (WSN) is an emerging technology has significant applications in several important fields like military, agriculture, healthcare, environmental, artificial intelligence and research. All these applications demands high quality data transmission from resource constraint WSNs. But interference is one of the severe problem in WSN which can degrade the quality data transmission. Various interference minimization techniques have been proposed but not results the expected degree of quality enhancement in WSNs. This research paper investigates various types of heterogeneous wireless sensor networks (HTWSN) deployment techniques, interference and its effects, existing interference minimization techniques with limitations. We propose interference minimization (IM) protocol for heterogeneous wireless sensor networks. IM protocol can efficiently minimize the interference and enhance the quality data transmission in WSN.

Research paper thumbnail of A Study on Speech Processing

Advances in Intelligent Systems and Computing, 2016

Speech is the most natural means of communication in human-to-human interactions. Automatic Speec... more Speech is the most natural means of communication in human-to-human interactions. Automatic Speech Recognition (ASR) is the application of technology in developing machines that can autonomously transcribe a speech into a text in the real-time. This paper presents a short review of ASR systems. Fundamentally, the design of speech recognition system involves three major processes such as feature extraction, acoustic modeling and classification. Consequently, emphasis is laid on describing essential principles of the various techniques employed in each of these processes. On the other hand, it also presents the milestones in the speech processing research to date.

Research paper thumbnail of Game Theory and Its Applications in Machine Learning

Advances in Intelligent Systems and Computing, 2016

Machine learning is a discipline that deals with the study of algorithms that can learn from the ... more Machine learning is a discipline that deals with the study of algorithms that can learn from the data. Typically, these algorithms run by generating a model built from the observed data, and then employ the generated model to predict and make decisions. Most of the problems in machine learning could be translated to multi-objective optimization problems where multiple objectives have to be optimized at the same time in the presence of two or more conflicting objectives. Mapping multi-optimization problems to game theory can give stable solutions. This paper presents an introduction of game theory and collects the survey on how game theory is applied to some of the machine learning problems.

Research paper thumbnail of A Fast Algorithm for HMM Training using Game Theory for Phoneme Recognition

International Journal of Computer Applications, 2015

Hidden Markov Models are widely used for modeling and predicting label sequences in ASR. In this ... more Hidden Markov Models are widely used for modeling and predicting label sequences in ASR. In this paper, a game-theoretic approach for Hidden Markov Model training that is superior in terms of time-complexity over Baum-Welch algorithm is introduced. Furthermore, accuracy of recognition using proposed algorithm is comparable with that of Baum-Welch algorithm.

Research paper thumbnail of Automatic Speech Segmentation and Recognition using Class-Specific Features

International Journal of Computer Applications, 2015

The class-specific automatic speech recognition systems construct an individual classifier for ea... more The class-specific automatic speech recognition systems construct an individual classifier for each class based on its own feature set, wherein the feature set for each class is selected such that it distinguishes that class from the other classes most accurately. Consequently, different feature set sequences must be fed into each of the classifiers, and the output of each of the classifiers must be combined to predict the actual class of the observation sequences. However, speech is continuous, and to be able to apply class-specific features, speech should be segmented and fed to the classifiers, which requires the identification of segmentation cues. This paper proposes a framework that jointly segments, and combines the output of the class-specific classifiers in the absence of any segmentation cues using a recursive formulation.

Research paper thumbnail of Game theoretic approach for automatic speech segmentation and recognition

2014 IEEE 28th Convention of Electrical & Electronics Engineers in Israel (IEEEI), 2014

In this paper, we propose a game-theoretic approach for automatic speech segmentation and recogni... more In this paper, we propose a game-theoretic approach for automatic speech segmentation and recognition. Speech segmentation and recognition are performed jointly by obtaining a Nash equilibrium solution. The proposed algorithm is tested on TIMIT database and compared against the Viterbi algorithm. Boundaries detected using the proposed algorithm are more accurate than the boundaries detected using the Viterbi algorithm.

Research paper thumbnail of Feature selection using game theory for phoneme based speech recognition

2014 International Conference on Contemporary Computing and Informatics (IC3I), 2014

Reduced feature set containing relevant features for identifying individual phonemes were obtaine... more Reduced feature set containing relevant features for identifying individual phonemes were obtained using two game-theoretic formulations. In one formulation feature selection algorithm tries to obtain features that maximize the accuracy of the classifier, and in another it obtains features that minimize the misclassification rate of the classifier. Experiments are run on the TIMIT database for generating classifiers using the reduced feature set obtained from our feature selection algorithms and compared against classifiers generated using all of the features. The results show that, classifiers generated using the reduced feature set out performed classifiers generated from all of the features. In addition, reduced feature sets obtained using proposed feature selection algorithms could significantly reduce storage and computational complexity without compromising on accuracy of classifiers.

Research paper thumbnail of Variable initial energy and unequal clustering (VEUC) based multicasting in WSN

2017 International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET), 2017

Multicast Communication plays an important role in most of the resource constrained networking en... more Multicast Communication plays an important role in most of the resource constrained networking environments such as Wireless Sensor Networks (WSN), Internet of Things (IOT). Communication in WSN is restricted by energy, computation and memory capabilities of sensor nodes. Designing an efficient routing algorithm to achieve communication between Stationary Base station (BS) and a cluster of sensor nodes in a specific region requires the base station to send individual messages to all sensor nodes. This approach consumes a large amount of energy and bandwidth. A variety of algorithms exist to address this issue by dividing the sensor nodes into clusters. Each cluster is monitored by a Cluster Head (CH), responsible for gathering and aggregating data to send the same to the BS. In this paper, we reviewed existing clustering techniques and propose an unequal clustering based scheme. This allows the BS to communicate a multicast message to cluster members as well as a cluster head to communicate with other cluster members. The results show that our approach improves network lifetime.

Research paper thumbnail of Energy-Efficient Neighbor Discovery Using Bacterial Foraging Optimization (BFO) Algorithm for Directional Wireless Sensor Networks

Lecture Notes in Electrical Engineering, 2021

In directional wireless sensor networks (WSN), the existing neighbor's discovery methods involve ... more In directional wireless sensor networks (WSN), the existing neighbor's discovery methods involve high latency and energy consumption, compared to the block design-based methods. Moreover, the duty cycle schedule of nodes has to be addressed to increase the network lifetime. In this paper, an energy-efficient collaborative neighbor discovery mechanism using the bacterial foraging optimization (BFO) algorithm is recommended. In this computation, each node with a directional antenna performs beamforming using BFOA with sector number and beam direction as the fitness function. In the end, appropriate active nodes with higher energy levels are selected from the neighbors during data transmission. The obtained results have shown that the recommended model minimizes power conservation and delay and enhances the lifetime of time of network activity.

Research paper thumbnail of Neighbor Nodes Discovery Schemes in a Wireless Sensor Network

Electronics and Communications Engineering, 2019

Research paper thumbnail of DEGSA-VMM: Dragonfly-based exponential gravitational search algorithm to VMM strategy for load balancing in cloud computing

Kybernetes, 2018

Purpose This paper aims to develop the Dragonfly-based exponential gravitational search algorithm... more Purpose This paper aims to develop the Dragonfly-based exponential gravitational search algorithm to VMM strategy for effective load balancing in cloud computing. Due to widespread growth of cloud users, load balancing is the essential criterion to deal with the overload and underload problems of the physical servers. DEGSA-VMM is introduced, which calculates the optimized position to perform the virtual machine migration (VMM). Design/methodology/approach This paper presents an algorithm Dragonfly-based exponential gravitational search algorithm (DEGSA) that is based on the VMM strategy to migrate the virtual machines of the overloaded physical machine to the other physical machine keeping in mind the energy, migration cost, load and quality of service (QoS) constraints. For effective migration, a fitness function is provided, which selects the best fit that possess minimum energy, cost, load and maximum QoS contributing toward the maximum energy utilization. Findings For the perfo...

Research paper thumbnail of ECC-Based Secure Group Communication in Energy-Efficient Unequal Clustered WSN (EEUC-ECC)

With an advent of the Internet of things (IoT), wireless sensor networks (WSNs) are gaining popul... more With an advent of the Internet of things (IoT), wireless sensor networks (WSNs) are gaining popularity in application areas like smart cities, body area sensor networks, industrial process control, and habitat and environment monitoring. Since these networks are exposed to various attacks like node compromise attack, DoS attacks, etc., the need for secured communication is evident. We present an updated survey on various secure group communication (SGC) schemes and evaluate their performance in terms of space and computational complexity. We also propose a novel technique for secure and scalable group communication that performs better compared with existing approaches.

Research paper thumbnail of Competitive Equilibrium Approach for Load Balancing a Computational Grid with Communication Delays

Computational grids interconnect hundreds of heterogeneous computing resources from geographicall... more Computational grids interconnect hundreds of heterogeneous computing resources from geographically remote sites, designed to meet the large demands of many users from scientific and business domains. A job initiated at one site can be executed by any of the computing resources. Therefore, response time of a job includes processing delay at the site of execution and communication delay for transferring the job from the site of initiation to the site of execution. Load balancing is allocation of jobs to available resources so as to optimize a given objective function. The objective can be achieving a system optimal solution, which tries to minimize the mean response time of all users or an individual optimal solution which tries to minimize each user’s response time. Previous works on load balancing either considered only system optimal objective or individual optimal objective. This paper introduces competitive equilibrium solution, a pricing mechanism for load balancing that indepen...

Research paper thumbnail of Competitive equilibrium approach for load balanicing a grid network

Research paper thumbnail of An Energy Efficient Node Scheduling based Congestion Control Scheme for WSN Multicasting

EAI Endorsed Transactions on Energy Web, 2018

Research paper thumbnail of A Study on Text Similarity Measures

Journal of Advanced Research in Dynamical and Control Systems, 2020

Research paper thumbnail of Competitive equilibrium based personal data market model

Indian Journal of Science and Technology, 2020

Background/Objectives: In a digital economy, with the increasing commercial value, personal data ... more Background/Objectives: In a digital economy, with the increasing commercial value, personal data is viewed as a commodity to be bought and sold. Data owners expect an appropriate compensation for trading off their privacy depending on how they value their privacy. Simultaneously, data consumers want to maximize their utility which is dependent on the value derived from the data. Consequently, a data market model that optimally recompenses data owners and maximizes the profits of data consumers is required. Methods: In this study, a data market model and pricing mechanism based on Fisher market model and competitive equilibrium are presented where the value derived from the data is calculated from information entropy. The proposed data model and the pricing mechanism jointly and simultaneously maximize the profit of data owners and the utility of data buyers. Findings: Experiments are conducted on adult data set to validate the efficacy of the proposed approach. Data owners are classified as risk averse, risk neutral, risk taking and privacy regarders. Subsequently, prices of data samples are adjusted to reach equilibrium as defined by the Fisher market model maximizing simultaneously and jointly the profit of data owners and the utility of data buyers. Applications: The proposed competitive equilibrium based personal data market model can be used to find equilibrium prices and bundles of data samples for each data buyer at these prices maximizing the utility of data buyer subject to his budgetary constraints and data requirements, and data owners' privacy preferences.

Research paper thumbnail of Exponential gravitational search algorithm-based VM migration strategy for load balancing in cloud computing

International Journal of Modeling, Simulation, and Scientific Computing, 2017

With the advancement in the science and technology, cloud computing has become a recent trend in ... more With the advancement in the science and technology, cloud computing has become a recent trend in environment with immense requirement of infrastructure and resources. Load balancing of cloud computing environments is an important matter of concern. The migration of the overloaded virtual machines (VMs) to the underloaded VM with optimized resource utilization is the effective way of the load balancing. In this paper, a new VM migration algorithm for the load balancing in the cloud is proposed. The migration algorithm proposed (EGSA-VMM) is based on exponential gravitational search algorithm which is the integration of gravitational search algorithm and exponential weighted moving average theory. In our approach, the migration is done based on the migration cost and QoS. The experimentation of proposed EGSA-based VM migration algorithm is compared with ACO and GSA. The simulation of experiments shows that the proposed EGSA-VMM algorithm achieves load balancing and reasonable resource...

Research paper thumbnail of New secure routing protocol with elliptic curve cryptography for military heterogeneous wireless sensor networks

Journal of Information and Optimization Sciences, 2017

Military Heterogeneous Wireless Sensor Network (MHTWSN) is an emerging technology. Due to its ext... more Military Heterogeneous Wireless Sensor Network (MHTWSN) is an emerging technology. Due to its extending services, MHTWSN is becoming a widely preferable network for border area monitoring and enemy object tracking in sensitive, unmanned areas. However, security is one of the key challenging issue in MHTWSN. Many routing protocols like AODV, AOMDV, I2MR, and IMCC have proposed for efficient routing establishment and data delivery, but these routing protocols results the poor performance when the network is under any security attack. So we have proposed a new secure multipath routing protocol

Research paper thumbnail of Multipath Interference Minimization in Heterogeneous Wireless Sensor Networks for Reliable Data Transfer

2016 International Conference on Computer and Communication Engineering (ICCCE), 2016

Multipath interference or route coupling is a severe issue in multipath routing schemes of wirele... more Multipath interference or route coupling is a severe issue in multipath routing schemes of wireless sensor networks. Due to that, quality of data transmission is impossible in wireless sensor network (WSN) or heterogeneous wireless sensor network (HTWSN). Multipath routing is having many features over single path routing, but it is highly prone to multipath interference. The existing multipath routing schemes can results poor network performance and low reliable data transmission due to the existence of multipath interference. So we propose interference minimization clustering multipath routing protocol (IMCMRP) for multipath interference minimization with energy efficiency and reliable data transmission. Simulation results shows that, IMCMRP protocol is achieved higher data delivery ratio 20% and 31%, better network life time 20 % and 42 %, lower latency up to 30 % and 60 %, higher throughput 30 % and 48 % than LIEMRO multipath routing and single path routing schemes.

Research paper thumbnail of Interference Minimization Protocol in Heterogeneous Wireless Sensor Networks for Military Applications

Smart Innovation, Systems and Technologies, 2016

Wireless sensor Network (WSN) is an emerging technology has significant applications in several i... more Wireless sensor Network (WSN) is an emerging technology has significant applications in several important fields like military, agriculture, healthcare, environmental, artificial intelligence and research. All these applications demands high quality data transmission from resource constraint WSNs. But interference is one of the severe problem in WSN which can degrade the quality data transmission. Various interference minimization techniques have been proposed but not results the expected degree of quality enhancement in WSNs. This research paper investigates various types of heterogeneous wireless sensor networks (HTWSN) deployment techniques, interference and its effects, existing interference minimization techniques with limitations. We propose interference minimization (IM) protocol for heterogeneous wireless sensor networks. IM protocol can efficiently minimize the interference and enhance the quality data transmission in WSN.

Research paper thumbnail of A Study on Speech Processing

Advances in Intelligent Systems and Computing, 2016

Speech is the most natural means of communication in human-to-human interactions. Automatic Speec... more Speech is the most natural means of communication in human-to-human interactions. Automatic Speech Recognition (ASR) is the application of technology in developing machines that can autonomously transcribe a speech into a text in the real-time. This paper presents a short review of ASR systems. Fundamentally, the design of speech recognition system involves three major processes such as feature extraction, acoustic modeling and classification. Consequently, emphasis is laid on describing essential principles of the various techniques employed in each of these processes. On the other hand, it also presents the milestones in the speech processing research to date.

Research paper thumbnail of Game Theory and Its Applications in Machine Learning

Advances in Intelligent Systems and Computing, 2016

Machine learning is a discipline that deals with the study of algorithms that can learn from the ... more Machine learning is a discipline that deals with the study of algorithms that can learn from the data. Typically, these algorithms run by generating a model built from the observed data, and then employ the generated model to predict and make decisions. Most of the problems in machine learning could be translated to multi-objective optimization problems where multiple objectives have to be optimized at the same time in the presence of two or more conflicting objectives. Mapping multi-optimization problems to game theory can give stable solutions. This paper presents an introduction of game theory and collects the survey on how game theory is applied to some of the machine learning problems.

Research paper thumbnail of A Fast Algorithm for HMM Training using Game Theory for Phoneme Recognition

International Journal of Computer Applications, 2015

Hidden Markov Models are widely used for modeling and predicting label sequences in ASR. In this ... more Hidden Markov Models are widely used for modeling and predicting label sequences in ASR. In this paper, a game-theoretic approach for Hidden Markov Model training that is superior in terms of time-complexity over Baum-Welch algorithm is introduced. Furthermore, accuracy of recognition using proposed algorithm is comparable with that of Baum-Welch algorithm.

Research paper thumbnail of Automatic Speech Segmentation and Recognition using Class-Specific Features

International Journal of Computer Applications, 2015

The class-specific automatic speech recognition systems construct an individual classifier for ea... more The class-specific automatic speech recognition systems construct an individual classifier for each class based on its own feature set, wherein the feature set for each class is selected such that it distinguishes that class from the other classes most accurately. Consequently, different feature set sequences must be fed into each of the classifiers, and the output of each of the classifiers must be combined to predict the actual class of the observation sequences. However, speech is continuous, and to be able to apply class-specific features, speech should be segmented and fed to the classifiers, which requires the identification of segmentation cues. This paper proposes a framework that jointly segments, and combines the output of the class-specific classifiers in the absence of any segmentation cues using a recursive formulation.

Research paper thumbnail of Game theoretic approach for automatic speech segmentation and recognition

2014 IEEE 28th Convention of Electrical & Electronics Engineers in Israel (IEEEI), 2014

In this paper, we propose a game-theoretic approach for automatic speech segmentation and recogni... more In this paper, we propose a game-theoretic approach for automatic speech segmentation and recognition. Speech segmentation and recognition are performed jointly by obtaining a Nash equilibrium solution. The proposed algorithm is tested on TIMIT database and compared against the Viterbi algorithm. Boundaries detected using the proposed algorithm are more accurate than the boundaries detected using the Viterbi algorithm.

Research paper thumbnail of Feature selection using game theory for phoneme based speech recognition

2014 International Conference on Contemporary Computing and Informatics (IC3I), 2014

Reduced feature set containing relevant features for identifying individual phonemes were obtaine... more Reduced feature set containing relevant features for identifying individual phonemes were obtained using two game-theoretic formulations. In one formulation feature selection algorithm tries to obtain features that maximize the accuracy of the classifier, and in another it obtains features that minimize the misclassification rate of the classifier. Experiments are run on the TIMIT database for generating classifiers using the reduced feature set obtained from our feature selection algorithms and compared against classifiers generated using all of the features. The results show that, classifiers generated using the reduced feature set out performed classifiers generated from all of the features. In addition, reduced feature sets obtained using proposed feature selection algorithms could significantly reduce storage and computational complexity without compromising on accuracy of classifiers.