Ramadevi Yellasiri - Academia.edu (original) (raw)

Papers by Ramadevi Yellasiri

Uncertainty is one of the main characteristics of knowledge handling systems. The central problem... more Uncertainty is one of the main characteristics of knowledge handling systems. The central problem in rough set theory is the construction of approximate operators. Many of the existing approaches are based on the binary relations of the axiomatic approaches. Based on the inclusion measures of fuzzy sets, a λ-weak fuzzy approximation space and a modified fuzzy rough set were defined. Rough set theory and fuzzy set theory deals with imperfect knowledge. Both theories can be combined into a more expressive framework for modelling and processing incomplete information systems. This paper demonstrates a new approach of determining the rough approximations of fuzzy concepts in fuzzy decision systems. A λ-weak fuzzy approximation space based on the hierarchical characteristic of fuzzy sets, determined by an inclusion measure is defined. It also demonstrates the ISλ – fuzzy Rough set that are based on a hybrid monotonic inclusion measure and similarity measure in a λ-weak fuzzy approximatio...

Proceedings of the 1st Amrita ACM-W Celebration on Women in Computing in India, 2010

... Classification Rama Devi Yellasiri Venu Gopal P Sai Prasad PSVS ... Fuzzy-rough sets for desc... more ... Classification Rama Devi Yellasiri Venu Gopal P Sai Prasad PSVS ... Fuzzy-rough sets for descriptive dimensionality reduction, Fuzzy Systems, FUZZ-IEEE'02: Proceedings of the 2002 IEEE International Conference, 29-34. [9] Venugopal, P., Rama Devi, Y., Sai Prasad PSVS. ...

A thesis submitted in partial fulfillment of the requirements for the award of the degree of

INASS, 2023

One of the most prominent indicators for the detection of breast cancer is a breast mass. In this... more One of the most prominent indicators for the detection of breast cancer is a breast mass. In this regard, effective mass segmentation for any type of mammography is crucial for improving breast cancer detection accuracy and lowering mortality. In order to pace up the process of mammogram segmentation for breast mass, an ABC3D (artificial bee colony based 3 dimensional) Otsu method is proposed in this paper. Firstly, convergence speed of bees in basic artificial bee colony (ABC) is improved by adopting the epsilon greedy method for scout bees. Secondly, proposed improved ABC method is paired with optimal 3D Otsu multilevel thresholding technique to get the better thresholding set for medical mammogram images. Epsilon greed based scout bee technique streamline the exploration-exploitation problem of standard ABC while searching for best threshold set in 3D space. The proposed ABC3D is tested on eight mammogram images collected from the authoritative and publicly available database mini MIAS (mammographic image analysis society). PSNR (peek signal to noise ration), SSIM (structural similarity index) and time cost are measured to record the effectiveness of ABC3D method. The results of experimentations indicate that the proposed ABC3D achieve superior segmentation results than the teaching learning ABC (TLABC), ABCDS (directed scout), gbest guided ABC (GABC), improved particle swarm optimization (IPSO) with 3D Otsu as objective functions.

JATIT, 2022

In recent past, artificial neural networks (ANN) have reaped improvements in the domain of medica... more In recent past, artificial neural networks (ANN) have reaped improvements in the domain of medical image processing by addressing many unmanageable problems. The initialized hyperparameters control ANN performance and selecting sensible hyperparameters by hand is time-consuming and tiresome. This study suggests a metaheuristic optimization of the fine-tuning hyperparameters approach to remedy this flaw. The method is then evaluated on mammography images to assess whether the mammogram contains cancer. In the proposed ANN model, a modified Artificial bee colony (ABC) optimization method is used to fine tune the hyperparameters, and it categorizes the tumors in the breast as benign or malignant in two-class case and normal, benign, and malignant in three-class case with an accuracy of 97.52% and 96.58% respectively. Hyperparameters to the neural network framework were assigned instantly with the help of ABC method with wrapped ANN as objective function. Manual search, Grid Search, Random Grid search, Bayes search are all cutting edge ANN hyperparameters methods. In addition to the mentioned, nature-inspired optimization methods such as PSO and GA have adopted for fine tuning parameters. Additionally, the suggested model's performance in classifying breast pictures was compared to that of the published hyperparameter technique using sizable datasets on breast cancer that were made accessible to the public.

Dimensional reduction has been a major problem in data mining problems. In many real time situati... more Dimensional reduction has been a major problem in data mining problems. In many real time situations, e.g. database applications and bioinformatics, there are far too many attributes to be handled by learning schemes, majority of them being redundant. Taking predominant attributes reduces the dimensions of the data, which in turn reduces the size of the hypothesis space, allowing classification algorithm to operate faster and more efficiently. The Rough Set (RS) theory is one such approach for dimension reduction. RS offers a simplified search for predominant attributes in datasets. Rough Set based Decision Tree(RDT) is constructed based on the predominant attributes. The comparative analysis with the existing decision tree algorithms was made to show that the intent of RDT is to improve efficiency, simplicity and generalization capability.

2021 International Conference on Forensics, Analytics, Big Data, Security (FABS), 2021

Alerts are tasks that continually monitor active queries to look for and report on specific event... more Alerts are tasks that continually monitor active queries to look for and report on specific events or conditions like system performance, security incidents, and threats for a system or network. Companies with an extensive IT infrastructure often deal with many alerts per day, varying from a routine host or network performance notifications to security incidents raised by Network Security Devices. With an increase in cyberattacks, Network Security Devices play a vital role in detecting critical incidents and threats. However, more often than not, these security incidents are frequently occurring low threat alerts. As the number of alerts skyrockets, it becomes increasingly tedious to sift through all the alerts generated and identify critical one. This may result in longer response time or overlooking important alerts, which is referred to as alert fatigue. Aiming to tackle this problem, our paper proposes a solution to reduce alert fatigue by identifying and highlighting anomalous alerts using Extended Isolation Forest, an isolation-based anomaly detection technique. Our model reduces the number of alerts received at the Security Operations Center (SOC) by 82.15%. The security analyst needs to monitor only 17.85% of the 50,000 total alerts received from the IDS.

Smart innovation, systems and technologies, 2022

Smart Intelligent Computing and Applications, Volume 1

Advances in Intelligent Systems and Computing, 2020

The subsequent generation of IoT devices must work on a multi-protocol architecture to facilitate... more The subsequent generation of IoT devices must work on a multi-protocol architecture to facilitate M2M communication along with endpoint user interfacing to solve the network infrastructure dependencies accompanied by redundant data flow overhead. An ideological solution is proposed to facilitate a change while cutting down infrastructure cost and enhancing the current setups through proper implementation of edge computation. End devices cooperate with each other along with providing GUI and Internet to handsets; monitoring sensor information as well as issuing control signals.

2019 Fifth International Conference on Image Information Processing (ICIIP), 2019

Internet of Things (IoT) is yet to be converted into a well-structured, self-configurable, dynami... more Internet of Things (IoT) is yet to be converted into a well-structured, self-configurable, dynamic global infrastructure. IoT focusses interoperable communication protocols where physical things having unique identifications and characteristics that are integrated into the data network. Various practices adopted by organizations in the study of IoT concentrate more about the working principles rather than enhancing their communication. MANET is a combination of decentralized clusters and hubs which are portable and has many applications in many territories. One fundamental preferred state for a decentralized system is that they are normally more powerful than concentrated centralized systems due to the multi-bounce design in which data is transferred. MANET additionally pursues some downsides in system execution but, it is undeniable that we ought to expect varieties in connections because of no fixed engineering correspondence in IoT. MANET application in IoT will help in proper self-configuration of devices for establishing dynamic communication.

Learning and Analytics in Intelligent Systems, 2019

IoT administrations are ordinarily conveyed of IoT as physically disconnected vertical arrangemen... more IoT administrations are ordinarily conveyed of IoT as physically disconnected vertical arrangements, in which all framework segments running from tangible gadgets to applications are firmly coupled for the prerequisites of each explicit venture. The productivity and versatility of such administration conveyance are naturally constrained, presenting noteworthy difficulties to IoT arrangement developers. In this context, we propose a novel SaaS structure that gives basic stage administrations to IoT arrangement suppliers to productively convey and constantly expand their administrations for DIY applications over HTTP with no investment required. This paper initially presents the IoT SaaS engineering, on which IoT arrangements can be conveyed as virtual verticals by utilizing figuring assets and middleware benefits on free cloud services. At that point we present the itemized instrument, usage of area intervention, which helps arrangement suppliers to productively give area explicit co...

2019 IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE), 2019

The Internet of Things (IoT) will grow seamlessly with advancements in data and communication tec... more The Internet of Things (IoT) will grow seamlessly with advancements in data and communication technologies leading to the deployment of trillions of end devices. Its application starts with a simple home automation to a very large scale industrial automation system. The trend is leading towards huge data generation requiring high processing power. In the near future, computing resources might not be sufficient for handling dynamic humongous data production. As the technology advances, microcontrollers or System-on-Chips (SoCs) used for IoT end devices are becoming cheaper and more powerful. Hence, there is a requirement of effectively making use of huge number of underutilized IoT of the future by allocating additional micro-tasks in parallel which would solve the upcoming needs of the technological trend.

International Journal of Computer Applications, Jan 31, 2012

Speech communication in Electronic Warfare (EW) environment should be resistant to interception, ... more Speech communication in Electronic Warfare (EW) environment should be resistant to interception, masquerade and tolerant to communication channel errors. In this paper, we described an algorithm which provides speech compression, strong encryption, error tolerance and speaker authentication features. This Robust Speech Coder (RSC) is backward compatible with the existing codecs with capability to opt for additional features as and when required.

Signal & Image Processing : An International Journal, 2015

The state-of-the-art Automatic Speech Recognition (ASR) systems lack the ability to identify spok... more The state-of-the-art Automatic Speech Recognition (ASR) systems lack the ability to identify spoken words if they have non-standard pronunciations. In this paper, we present a new classification algorithm to identify pronunciation variants. It uses Dynamic Phone Warping (DPW) technique to compute the pronunciation-by-pronunciation phonetic distance and a threshold critical distance criterion for the classification. The proposed method consists of two steps; a training step to estimate a critical distance parameter using transcribed data and in the second step, use this critical distance criterion to classify the input utterances into the pronunciation variants and OOV words. The algorithm is implemented using Java language. The classifier is trained on data sets from TIMIT speech corpus and CMU pronunciation dictionary. The confusion matrix and precision, recall and accuracy performance metrics are used for the performance evaluation. Experimental results show significant performance improvement over the existing classifiers.

International Journal of Computer and Electrical Engineering, 2011

Fuzzy rough data reduction algorithm proposed in [1] is not convergent on higher dimensional data... more Fuzzy rough data reduction algorithm proposed in [1] is not convergent on higher dimensional data due to its computational complexity increases exponentially as the number of input attributes and fuzzy sets increase. This paper shows how singular value decomposition can be used as a useful preprocessing method in order to achieve fuzzy rough reduct convergence on higher dimensional datasets. Eight datasets from UCI repository have been taken for the experimentation.

Classification of voluminous protein data based on the structural and functional properties is a ... more Classification of voluminous protein data based on the structural and functional properties is a challenging task for researchers in bioinformatics field. In this paper a faster, accurate and efficient classification tool Rough Set Protein Classifier has been developed which has a classification accuracy of 97.7%. It is a hybridized tool comprising Sequence Arithmetic, Rough Set Theory and Concept Lattice. It reduces the domain search space to 9% without losing the potentiality of classification of proteins.

Where human beings can easily learn and adopt pronunciation variations, machines need training be... more Where human beings can easily learn and adopt pronunciation variations, machines need training before put into use. Also humans keep minimum vocabulary and their pronunciation variations are stored in front-end of their memory for ready reference, while machines keep the entire pronunciation dictionary for ready reference. Supervised methods are used for preparation of pronunciation dictionaries which take large amounts of manual effort, cost, time and are not suitable for real time use. This paper presents an unsupervised adaptation model for building agile and dynamic pronunciation dictionaries online. These methods mimic human approach in learning the new pronunciations in real time. A new algorithm for measuring sound distances called Dynamic Phone Warping is presented and tested. Performance of the system is measured using an adaptation model and the precision metrics is found to be better than 86 percent.

Computer Science & Information Technology, 2018

Human beings generate different speech waveforms while speaking the same word at different times.... more Human beings generate different speech waveforms while speaking the same word at different times. Also, different human beings have different accents and generate significantly varying speech waveforms for the same word. There is a need to measure the distances between various words which facilitate preparation of pronunciation dictionaries. A new algorithm called Dynamic Phone Warping (DPW) is presented in this paper. It uses dynamic programming technique for global alignment and shortest distance measurements. The DPW algorithm can be used to enhance the pronunciation dictionaries of the well-known languages like English or to build pronunciation dictionaries to the less known sparse languages. The precision measurement experiments show 88.9% accuracy.

IETE Journal of Research, 2016

ABSTRACT There is a large gap between the capabilities of the human beings and the automatic spee... more ABSTRACT There is a large gap between the capabilities of the human beings and the automatic speech recognition (ASR) systems in recognizing pronunciation variations. ASR systems learn from labelled speech corpus, whereas the humans use “Everyday Speech” for adapting pronunciation variability. Labelling huge speech corpus in real time is impracticable, expensive, and time-consuming. In this paper, we present an algorithm using unsupervised learning techniques for adapting the easily available “Everyday Speech”. The algorithm is implemented using Java. The data sets are extracted from CMUDICT pronunciation directory, TIMIT database, and “The Hindu” daily newspaper. The results have shown a significant improvement in word error rate (WER) measurements over the existing ASR system. The addition of dynamic pronunciation model enables the ASR system to learn from the unlabelled “Everyday Speech” and makes it inexpensive and fast.

Uncertainty is one of the main characteristics of knowledge handling systems. The central problem... more Uncertainty is one of the main characteristics of knowledge handling systems. The central problem in rough set theory is the construction of approximate operators. Many of the existing approaches are based on the binary relations of the axiomatic approaches. Based on the inclusion measures of fuzzy sets, a λ-weak fuzzy approximation space and a modified fuzzy rough set were defined. Rough set theory and fuzzy set theory deals with imperfect knowledge. Both theories can be combined into a more expressive framework for modelling and processing incomplete information systems. This paper demonstrates a new approach of determining the rough approximations of fuzzy concepts in fuzzy decision systems. A λ-weak fuzzy approximation space based on the hierarchical characteristic of fuzzy sets, determined by an inclusion measure is defined. It also demonstrates the ISλ – fuzzy Rough set that are based on a hybrid monotonic inclusion measure and similarity measure in a λ-weak fuzzy approximatio...

Proceedings of the 1st Amrita ACM-W Celebration on Women in Computing in India, 2010

... Classification Rama Devi Yellasiri Venu Gopal P Sai Prasad PSVS ... Fuzzy-rough sets for desc... more ... Classification Rama Devi Yellasiri Venu Gopal P Sai Prasad PSVS ... Fuzzy-rough sets for descriptive dimensionality reduction, Fuzzy Systems, FUZZ-IEEE'02: Proceedings of the 2002 IEEE International Conference, 29-34. [9] Venugopal, P., Rama Devi, Y., Sai Prasad PSVS. ...

A thesis submitted in partial fulfillment of the requirements for the award of the degree of

INASS, 2023

One of the most prominent indicators for the detection of breast cancer is a breast mass. In this... more One of the most prominent indicators for the detection of breast cancer is a breast mass. In this regard, effective mass segmentation for any type of mammography is crucial for improving breast cancer detection accuracy and lowering mortality. In order to pace up the process of mammogram segmentation for breast mass, an ABC3D (artificial bee colony based 3 dimensional) Otsu method is proposed in this paper. Firstly, convergence speed of bees in basic artificial bee colony (ABC) is improved by adopting the epsilon greedy method for scout bees. Secondly, proposed improved ABC method is paired with optimal 3D Otsu multilevel thresholding technique to get the better thresholding set for medical mammogram images. Epsilon greed based scout bee technique streamline the exploration-exploitation problem of standard ABC while searching for best threshold set in 3D space. The proposed ABC3D is tested on eight mammogram images collected from the authoritative and publicly available database mini MIAS (mammographic image analysis society). PSNR (peek signal to noise ration), SSIM (structural similarity index) and time cost are measured to record the effectiveness of ABC3D method. The results of experimentations indicate that the proposed ABC3D achieve superior segmentation results than the teaching learning ABC (TLABC), ABCDS (directed scout), gbest guided ABC (GABC), improved particle swarm optimization (IPSO) with 3D Otsu as objective functions.

JATIT, 2022

In recent past, artificial neural networks (ANN) have reaped improvements in the domain of medica... more In recent past, artificial neural networks (ANN) have reaped improvements in the domain of medical image processing by addressing many unmanageable problems. The initialized hyperparameters control ANN performance and selecting sensible hyperparameters by hand is time-consuming and tiresome. This study suggests a metaheuristic optimization of the fine-tuning hyperparameters approach to remedy this flaw. The method is then evaluated on mammography images to assess whether the mammogram contains cancer. In the proposed ANN model, a modified Artificial bee colony (ABC) optimization method is used to fine tune the hyperparameters, and it categorizes the tumors in the breast as benign or malignant in two-class case and normal, benign, and malignant in three-class case with an accuracy of 97.52% and 96.58% respectively. Hyperparameters to the neural network framework were assigned instantly with the help of ABC method with wrapped ANN as objective function. Manual search, Grid Search, Random Grid search, Bayes search are all cutting edge ANN hyperparameters methods. In addition to the mentioned, nature-inspired optimization methods such as PSO and GA have adopted for fine tuning parameters. Additionally, the suggested model's performance in classifying breast pictures was compared to that of the published hyperparameter technique using sizable datasets on breast cancer that were made accessible to the public.

Dimensional reduction has been a major problem in data mining problems. In many real time situati... more Dimensional reduction has been a major problem in data mining problems. In many real time situations, e.g. database applications and bioinformatics, there are far too many attributes to be handled by learning schemes, majority of them being redundant. Taking predominant attributes reduces the dimensions of the data, which in turn reduces the size of the hypothesis space, allowing classification algorithm to operate faster and more efficiently. The Rough Set (RS) theory is one such approach for dimension reduction. RS offers a simplified search for predominant attributes in datasets. Rough Set based Decision Tree(RDT) is constructed based on the predominant attributes. The comparative analysis with the existing decision tree algorithms was made to show that the intent of RDT is to improve efficiency, simplicity and generalization capability.

2021 International Conference on Forensics, Analytics, Big Data, Security (FABS), 2021

Alerts are tasks that continually monitor active queries to look for and report on specific event... more Alerts are tasks that continually monitor active queries to look for and report on specific events or conditions like system performance, security incidents, and threats for a system or network. Companies with an extensive IT infrastructure often deal with many alerts per day, varying from a routine host or network performance notifications to security incidents raised by Network Security Devices. With an increase in cyberattacks, Network Security Devices play a vital role in detecting critical incidents and threats. However, more often than not, these security incidents are frequently occurring low threat alerts. As the number of alerts skyrockets, it becomes increasingly tedious to sift through all the alerts generated and identify critical one. This may result in longer response time or overlooking important alerts, which is referred to as alert fatigue. Aiming to tackle this problem, our paper proposes a solution to reduce alert fatigue by identifying and highlighting anomalous alerts using Extended Isolation Forest, an isolation-based anomaly detection technique. Our model reduces the number of alerts received at the Security Operations Center (SOC) by 82.15%. The security analyst needs to monitor only 17.85% of the 50,000 total alerts received from the IDS.

Smart innovation, systems and technologies, 2022

Smart Intelligent Computing and Applications, Volume 1

Advances in Intelligent Systems and Computing, 2020

The subsequent generation of IoT devices must work on a multi-protocol architecture to facilitate... more The subsequent generation of IoT devices must work on a multi-protocol architecture to facilitate M2M communication along with endpoint user interfacing to solve the network infrastructure dependencies accompanied by redundant data flow overhead. An ideological solution is proposed to facilitate a change while cutting down infrastructure cost and enhancing the current setups through proper implementation of edge computation. End devices cooperate with each other along with providing GUI and Internet to handsets; monitoring sensor information as well as issuing control signals.

2019 Fifth International Conference on Image Information Processing (ICIIP), 2019

Internet of Things (IoT) is yet to be converted into a well-structured, self-configurable, dynami... more Internet of Things (IoT) is yet to be converted into a well-structured, self-configurable, dynamic global infrastructure. IoT focusses interoperable communication protocols where physical things having unique identifications and characteristics that are integrated into the data network. Various practices adopted by organizations in the study of IoT concentrate more about the working principles rather than enhancing their communication. MANET is a combination of decentralized clusters and hubs which are portable and has many applications in many territories. One fundamental preferred state for a decentralized system is that they are normally more powerful than concentrated centralized systems due to the multi-bounce design in which data is transferred. MANET additionally pursues some downsides in system execution but, it is undeniable that we ought to expect varieties in connections because of no fixed engineering correspondence in IoT. MANET application in IoT will help in proper self-configuration of devices for establishing dynamic communication.

Learning and Analytics in Intelligent Systems, 2019

IoT administrations are ordinarily conveyed of IoT as physically disconnected vertical arrangemen... more IoT administrations are ordinarily conveyed of IoT as physically disconnected vertical arrangements, in which all framework segments running from tangible gadgets to applications are firmly coupled for the prerequisites of each explicit venture. The productivity and versatility of such administration conveyance are naturally constrained, presenting noteworthy difficulties to IoT arrangement developers. In this context, we propose a novel SaaS structure that gives basic stage administrations to IoT arrangement suppliers to productively convey and constantly expand their administrations for DIY applications over HTTP with no investment required. This paper initially presents the IoT SaaS engineering, on which IoT arrangements can be conveyed as virtual verticals by utilizing figuring assets and middleware benefits on free cloud services. At that point we present the itemized instrument, usage of area intervention, which helps arrangement suppliers to productively give area explicit co...

2019 IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE), 2019

The Internet of Things (IoT) will grow seamlessly with advancements in data and communication tec... more The Internet of Things (IoT) will grow seamlessly with advancements in data and communication technologies leading to the deployment of trillions of end devices. Its application starts with a simple home automation to a very large scale industrial automation system. The trend is leading towards huge data generation requiring high processing power. In the near future, computing resources might not be sufficient for handling dynamic humongous data production. As the technology advances, microcontrollers or System-on-Chips (SoCs) used for IoT end devices are becoming cheaper and more powerful. Hence, there is a requirement of effectively making use of huge number of underutilized IoT of the future by allocating additional micro-tasks in parallel which would solve the upcoming needs of the technological trend.

International Journal of Computer Applications, Jan 31, 2012

Speech communication in Electronic Warfare (EW) environment should be resistant to interception, ... more Speech communication in Electronic Warfare (EW) environment should be resistant to interception, masquerade and tolerant to communication channel errors. In this paper, we described an algorithm which provides speech compression, strong encryption, error tolerance and speaker authentication features. This Robust Speech Coder (RSC) is backward compatible with the existing codecs with capability to opt for additional features as and when required.

Signal & Image Processing : An International Journal, 2015

The state-of-the-art Automatic Speech Recognition (ASR) systems lack the ability to identify spok... more The state-of-the-art Automatic Speech Recognition (ASR) systems lack the ability to identify spoken words if they have non-standard pronunciations. In this paper, we present a new classification algorithm to identify pronunciation variants. It uses Dynamic Phone Warping (DPW) technique to compute the pronunciation-by-pronunciation phonetic distance and a threshold critical distance criterion for the classification. The proposed method consists of two steps; a training step to estimate a critical distance parameter using transcribed data and in the second step, use this critical distance criterion to classify the input utterances into the pronunciation variants and OOV words. The algorithm is implemented using Java language. The classifier is trained on data sets from TIMIT speech corpus and CMU pronunciation dictionary. The confusion matrix and precision, recall and accuracy performance metrics are used for the performance evaluation. Experimental results show significant performance improvement over the existing classifiers.

International Journal of Computer and Electrical Engineering, 2011

Fuzzy rough data reduction algorithm proposed in [1] is not convergent on higher dimensional data... more Fuzzy rough data reduction algorithm proposed in [1] is not convergent on higher dimensional data due to its computational complexity increases exponentially as the number of input attributes and fuzzy sets increase. This paper shows how singular value decomposition can be used as a useful preprocessing method in order to achieve fuzzy rough reduct convergence on higher dimensional datasets. Eight datasets from UCI repository have been taken for the experimentation.

Classification of voluminous protein data based on the structural and functional properties is a ... more Classification of voluminous protein data based on the structural and functional properties is a challenging task for researchers in bioinformatics field. In this paper a faster, accurate and efficient classification tool Rough Set Protein Classifier has been developed which has a classification accuracy of 97.7%. It is a hybridized tool comprising Sequence Arithmetic, Rough Set Theory and Concept Lattice. It reduces the domain search space to 9% without losing the potentiality of classification of proteins.

Where human beings can easily learn and adopt pronunciation variations, machines need training be... more Where human beings can easily learn and adopt pronunciation variations, machines need training before put into use. Also humans keep minimum vocabulary and their pronunciation variations are stored in front-end of their memory for ready reference, while machines keep the entire pronunciation dictionary for ready reference. Supervised methods are used for preparation of pronunciation dictionaries which take large amounts of manual effort, cost, time and are not suitable for real time use. This paper presents an unsupervised adaptation model for building agile and dynamic pronunciation dictionaries online. These methods mimic human approach in learning the new pronunciations in real time. A new algorithm for measuring sound distances called Dynamic Phone Warping is presented and tested. Performance of the system is measured using an adaptation model and the precision metrics is found to be better than 86 percent.

Computer Science & Information Technology, 2018

Human beings generate different speech waveforms while speaking the same word at different times.... more Human beings generate different speech waveforms while speaking the same word at different times. Also, different human beings have different accents and generate significantly varying speech waveforms for the same word. There is a need to measure the distances between various words which facilitate preparation of pronunciation dictionaries. A new algorithm called Dynamic Phone Warping (DPW) is presented in this paper. It uses dynamic programming technique for global alignment and shortest distance measurements. The DPW algorithm can be used to enhance the pronunciation dictionaries of the well-known languages like English or to build pronunciation dictionaries to the less known sparse languages. The precision measurement experiments show 88.9% accuracy.

IETE Journal of Research, 2016

ABSTRACT There is a large gap between the capabilities of the human beings and the automatic spee... more ABSTRACT There is a large gap between the capabilities of the human beings and the automatic speech recognition (ASR) systems in recognizing pronunciation variations. ASR systems learn from labelled speech corpus, whereas the humans use “Everyday Speech” for adapting pronunciation variability. Labelling huge speech corpus in real time is impracticable, expensive, and time-consuming. In this paper, we present an algorithm using unsupervised learning techniques for adapting the easily available “Everyday Speech”. The algorithm is implemented using Java. The data sets are extracted from CMUDICT pronunciation directory, TIMIT database, and “The Hindu” daily newspaper. The results have shown a significant improvement in word error rate (WER) measurements over the existing ASR system. The addition of dynamic pronunciation model enables the ASR system to learn from the unlabelled “Everyday Speech” and makes it inexpensive and fast.

AES, 2022

Adopting Nature inspired optimization algorithms for image processing is on the double growth in ... more Adopting Nature inspired optimization algorithms for image processing is on the double growth in the last decade. Artificial Bee Colony (ABC) approach is highly potential nature inspired optimization method mimics the bee's foraging behaviour. Moreover, the popularity of classification and artificial intelligence in different fields leads to the employment of ABC algorithm in upsurge. Notably, Early detection of breast cancer through digital mammogram images is essential as it is the one the most common cause of humankind cancer deaths. the aim of this comprehensive survey was to methodically analyse the effectiveness of using ABC algorithm in medical image enhancement, segmentation, and classification. This study firstly gives introduction of ABC algorithm and its basic mathematical and biological principles and operations respectively. Furthermore, this academic study summarizes the ABC applications on image segmentation techniques like Otsu and image classification approaches like Artificial Neural Networks (ANN). Finally, this far-reaching study come up with the challenges while exercising ABC algorithm in medical image processing, especially in mammogram images.