Chandrasekhar Kambhampati - Academia.edu (original) (raw)
Papers by Chandrasekhar Kambhampati
Control, 1988. CONTROL 88., International Conference on, Apr 13, 1988
There are no difficulties in the model based steady-state optimising control problem if the model... more There are no difficulties in the model based steady-state optimising control problem if the model is a faithful representation of the process. However, in general, when model-reality differences exist, the results obtained tend to be sub-optimal. In order to obtain optimal values a technique which used integrated system optimisation and parameter estimation (ISOPE) was proposed by Roberts (1979).
This paper documents the practical application of a genetic algorithm based process controller. A... more This paper documents the practical application of a genetic algorithm based process controller. As opposed to using the genetic algorithm to optimise the parameters of a conventional controller, this technique uses the generalised predictive control formulation to evolve a control strategy based upon a performance index. The intention is to use a radial basis function neural network as the process model, and hence to provide nonlinear control.
... [13] Servin M., Cuevas FJ,(1993) A new kind of neu-ral network based on radial basis functio... more ... [13] Servin M., Cuevas FJ,(1993) A new kind of neu-ral network based on radial basis functions, Revista Mexicana de Fisica 39, No.2. pp.235-249. [14] Shannon CE Weaver W. (1949), A Mathematical Theory of Communication, University of Illinois Press, Champaign, Illinois. ...
J, Dec 25, 2018
Aβ accumulation has been discovered to form large, relatively cation-permeable channels in the pl... more Aβ accumulation has been discovered to form large, relatively cation-permeable channels in the plasma membrane of a neuron. These channel formations in the membranes of a neuron could cause cell depolarisation, sodium and potassium dysregulation, depletion of neural energy stores and other types of cellular dysfunction. This study shows that the build-up of amyloid beta (Aβ) depositions during the onset of Alzheimer's disease has profound effects on the activity of the local community of neurons in the central nervous system. These effects can include enhanced neural activity, spontaneous epileptiform activity and incidence of epileptic seizures. From the results in this area, it can be seen that the neurodegeneration observed in Alzheimer's disease has been associated with the increase of toxicity of Aβ depositions. In this research paper, we examined this hypothesis in light of a computational model of a neuron.
Lecture notes in networks and systems, Aug 23, 2017
The missing data issue is a fundamental challenge in terms of analyses and classification of data... more The missing data issue is a fundamental challenge in terms of analyses and classification of data. The classification performance of incomplete data could be affected and produce different accuracy results compared with complete data. In this work we compare six scalable imputation methods, implemented on a Heart Failure dataset. The comparison is done by the performance metrics of three different classification methods namely J48, REPTree, and Random Forest. The aim of the research is to find a classifier that achieves best performance results after imputing the missing data using different imputation methods. The results show that in general, the Random Forest classification achieves the best results in comparison to the decision tree J48 and REP Tree. Furthermore, the performance of classification improved when imputing the missing values by concept most common (CMC) and support vector machine (SVM).
2nd IET International Conference on Technologies for Active and Assisted Living (TechAAL 2016), 2016
Although randomised controlled studies have demonstrated that home tele monitoring might improve ... more Although randomised controlled studies have demonstrated that home tele monitoring might improve outcomes for patients being discharged from hospital following an admission for heart failure, it is not known whether the effect is seen in patients encountered in daily practice. We investigated the impact of Home Telehealth Monitoring (HTM) using a dedicated heart failure database in Hull, UK. We used propensity matching to compare outcomes between patients receiving HTM and those not. After matching there were 202 patients (26% Female, 68.3 ± 12.5 years). The primary endpoint was mortality at 1 year. Patients who received usual care had a greater risk of 1 year mortality (HR: 3.20, 95% CI: 1.40 - 7.28, P = 0.006) and a greater risk of death at 3 years (HR: 1.75, 95% CI: 1.05- 2.90, P = 0.03). There was no difference between the groups in the composite endpoint of hospitalisation or death within one year (HR: 0.74, 95% CI: 0.51- 1.07, P = 0.11). Patients using HTM spent 96% of the available days alive and out of hospital compared with 87% of the usual care group. HTM thus improves outcomes in realworld patients discharged from hospital following an admission for heart failure.
Proceedings of SAI Intelligent Systems Conference (IntelliSys) 2016, 2017
The missing data issue is a fundamental challenge in terms of analyses and classification of data... more The missing data issue is a fundamental challenge in terms of analyses and classification of data. The classification performance of incomplete data could be affected and produce different accuracy results compared with complete data. In this work we compare six scalable imputation methods, implemented on a Heart Failure dataset. The comparison is done by the performance metrics of three different classification methods namely J48, REPTree, and Random Forest. The aim of the research is to find a classifier that achieves best performance results after imputing the missing data using different imputation methods. The results show that in general, the Random Forest classification achieves the best results in comparison to the decision tree J48 and REP Tree. Furthermore, the performance of classification improved when imputing the missing values by concept most common (CMC) and support vector machine (SVM).
J, 2019
Most neurodegenerative diseases (NDD) are a result of changes in the chemical composition of neur... more Most neurodegenerative diseases (NDD) are a result of changes in the chemical composition of neurons. For example, Alzheimer’s disease (AD) is the product of Aβ peptide deposition which results in changes in the ion concentration. These changes in ion concentration affect the responses of the neuron to stimuli and often result in inducing excessive excitation or inhibition. This paper investigates the dynamics of a single neuron as ion changes occur. These changes are incorporated using the Nernst equation. Within the central and peripheral nervous system, signals and hence rhythms, are propagated through the coupling of the neurons. It was found that under certain conditions the coupling strength between two neurons could mitigate changes in ion concentration. By defining the state of perfect synchrony, it was shown that the effect of ion imbalance in coupled neurons was reduced while in uncoupled neurons these changes had a more significant impact on the neuronal behavior.
IFAC Proceedings Volumes, 1999
Proceedings of the International Joint Conference on Neural Networks, 2003.
Cluster Analysis is traditionally an unsupervised data reduction technique. However, hy unifying ... more Cluster Analysis is traditionally an unsupervised data reduction technique. However, hy unifying ideas from both Cluster Analysis and Rough Set Theory, the inherent structure of a data set can be expressed as a rule set which, in turn, can be modified in a supervised manner to obtain a decision .rule set with minimal ambiguity. This process of encasing knowledge extraction in an algorithmic framework results in an optimal process for decision support.
International Conference on Control '94, 1994
ABSTRACT Three types of recurrent network configurations have been proposed since they enable ade... more ABSTRACT Three types of recurrent network configurations have been proposed since they enable adequate description of temporal behaviour. The concept of relative order has been introduced so as to provide a framework for analysing such network configurations. It has been demonstrated that, excluding pathologies, each configuration is of relative order unity. Therefore it follows that an inverse exists for each type of configuration (Tsinias and Kalouptsidis, 1983; Hirschorn, 1979).
Biological Cybernetics, 2003
Fifth International Conference on FACTORY 2000 - The Technology Exploitation Process, 1997
The modelling and control of systems with linear mathematics is a well established field. However... more The modelling and control of systems with linear mathematics is a well established field. However, for applications in the real world, the development of nonlinear modelling and control techniques continues. Following the successful integration of a nonlinear modelling capability, using neural networks, into the Connoisseur/sup TM/ advanced control package, this paper describes the development and integration of a genetic algorithm (GA) based nonlinear controller. The control problem is outlined, an overview of genetic algorithms is given, and the use of a genetic algorithm as a controller is explained. Example results are shown, with comments about the various factors that affect the performance of the controller, and the practical implementation of the controller.
HAL (Le Centre pour la Communication Scientifique Directe), 2006
IFAC Proceedings Volumes, 2006
The last decade has seen a rapid increase and convergence in the use of (a) digital network techn... more The last decade has seen a rapid increase and convergence in the use of (a) digital network technology, (b) embedded systems and (c) fault analysis and on-line diagnosis. This has resulted in an increase in complexity of the resultant system which must not only ensure performance but also monitor and compensate for faults and hence avoid total failure. The advances made have facilitated a "plug-and-play" (PnP) characteristic of networked embedded systems. This paper provides a framework for the analysis and development of strategies which enable a fault-tolerant PnP feature.
This paper describes the development of a genetic algorithm based nonlinear controller. It builds... more This paper describes the development of a genetic algorithm based nonlinear controller. It builds on the successful integration of the modelling capability of an artificial neural network approach within the advanced control package Connoisseur/sup TM/. The case for the long standing need for a generalised nonlinear controller for handling practical nonlinear and time varying systems is made. The motivation in terms of achieving tighter control, leading to increased efficiency and profitability are stressed. Existing techniques of gain scheduling and multiple models are briefly discussed, as well as their limitations. In the approach adopted in this paper the genetic algorithm based controller is employed to search for an optimal set of control outputs to minimise a given performance index. The paper gives examples of simulation studies and comments of the various factors that affect the performance of the controller and the practical implementation of the controller.
AI, Computer Science and Robotics Technology
This article provides an optimisation method using a Genetic Algorithm approach to apply feature ... more This article provides an optimisation method using a Genetic Algorithm approach to apply feature selection techniques for large data sets to improve accuracy. This is achieved through improved classification, a reduced number of features, and furthermore it aids in interpreting the model. A clinical dataset, based on heart failure, is used to illustrate the nature of the problem and to show the effectiveness of the techniques developed. Clinical datasets are sometimes characterised as having many variables. For instance, blood biochemistry data has more than 60 variables that have led to complexities in developing predictions of outcomes using machine-learning and other algorithms. Hence, techniques to make them more tractable are required. Genetic Algorithms can provide an efficient and low numerically complex method for effectively selecting features. In this paper, a way to estimate the number of required variables is presented, and a genetic algorithm is used in a “wrapper” form...
Control, 1988. CONTROL 88., International Conference on, Apr 13, 1988
There are no difficulties in the model based steady-state optimising control problem if the model... more There are no difficulties in the model based steady-state optimising control problem if the model is a faithful representation of the process. However, in general, when model-reality differences exist, the results obtained tend to be sub-optimal. In order to obtain optimal values a technique which used integrated system optimisation and parameter estimation (ISOPE) was proposed by Roberts (1979).
This paper documents the practical application of a genetic algorithm based process controller. A... more This paper documents the practical application of a genetic algorithm based process controller. As opposed to using the genetic algorithm to optimise the parameters of a conventional controller, this technique uses the generalised predictive control formulation to evolve a control strategy based upon a performance index. The intention is to use a radial basis function neural network as the process model, and hence to provide nonlinear control.
... [13] Servin M., Cuevas FJ,(1993) A new kind of neu-ral network based on radial basis functio... more ... [13] Servin M., Cuevas FJ,(1993) A new kind of neu-ral network based on radial basis functions, Revista Mexicana de Fisica 39, No.2. pp.235-249. [14] Shannon CE Weaver W. (1949), A Mathematical Theory of Communication, University of Illinois Press, Champaign, Illinois. ...
J, Dec 25, 2018
Aβ accumulation has been discovered to form large, relatively cation-permeable channels in the pl... more Aβ accumulation has been discovered to form large, relatively cation-permeable channels in the plasma membrane of a neuron. These channel formations in the membranes of a neuron could cause cell depolarisation, sodium and potassium dysregulation, depletion of neural energy stores and other types of cellular dysfunction. This study shows that the build-up of amyloid beta (Aβ) depositions during the onset of Alzheimer's disease has profound effects on the activity of the local community of neurons in the central nervous system. These effects can include enhanced neural activity, spontaneous epileptiform activity and incidence of epileptic seizures. From the results in this area, it can be seen that the neurodegeneration observed in Alzheimer's disease has been associated with the increase of toxicity of Aβ depositions. In this research paper, we examined this hypothesis in light of a computational model of a neuron.
Lecture notes in networks and systems, Aug 23, 2017
The missing data issue is a fundamental challenge in terms of analyses and classification of data... more The missing data issue is a fundamental challenge in terms of analyses and classification of data. The classification performance of incomplete data could be affected and produce different accuracy results compared with complete data. In this work we compare six scalable imputation methods, implemented on a Heart Failure dataset. The comparison is done by the performance metrics of three different classification methods namely J48, REPTree, and Random Forest. The aim of the research is to find a classifier that achieves best performance results after imputing the missing data using different imputation methods. The results show that in general, the Random Forest classification achieves the best results in comparison to the decision tree J48 and REP Tree. Furthermore, the performance of classification improved when imputing the missing values by concept most common (CMC) and support vector machine (SVM).
2nd IET International Conference on Technologies for Active and Assisted Living (TechAAL 2016), 2016
Although randomised controlled studies have demonstrated that home tele monitoring might improve ... more Although randomised controlled studies have demonstrated that home tele monitoring might improve outcomes for patients being discharged from hospital following an admission for heart failure, it is not known whether the effect is seen in patients encountered in daily practice. We investigated the impact of Home Telehealth Monitoring (HTM) using a dedicated heart failure database in Hull, UK. We used propensity matching to compare outcomes between patients receiving HTM and those not. After matching there were 202 patients (26% Female, 68.3 ± 12.5 years). The primary endpoint was mortality at 1 year. Patients who received usual care had a greater risk of 1 year mortality (HR: 3.20, 95% CI: 1.40 - 7.28, P = 0.006) and a greater risk of death at 3 years (HR: 1.75, 95% CI: 1.05- 2.90, P = 0.03). There was no difference between the groups in the composite endpoint of hospitalisation or death within one year (HR: 0.74, 95% CI: 0.51- 1.07, P = 0.11). Patients using HTM spent 96% of the available days alive and out of hospital compared with 87% of the usual care group. HTM thus improves outcomes in realworld patients discharged from hospital following an admission for heart failure.
Proceedings of SAI Intelligent Systems Conference (IntelliSys) 2016, 2017
The missing data issue is a fundamental challenge in terms of analyses and classification of data... more The missing data issue is a fundamental challenge in terms of analyses and classification of data. The classification performance of incomplete data could be affected and produce different accuracy results compared with complete data. In this work we compare six scalable imputation methods, implemented on a Heart Failure dataset. The comparison is done by the performance metrics of three different classification methods namely J48, REPTree, and Random Forest. The aim of the research is to find a classifier that achieves best performance results after imputing the missing data using different imputation methods. The results show that in general, the Random Forest classification achieves the best results in comparison to the decision tree J48 and REP Tree. Furthermore, the performance of classification improved when imputing the missing values by concept most common (CMC) and support vector machine (SVM).
J, 2019
Most neurodegenerative diseases (NDD) are a result of changes in the chemical composition of neur... more Most neurodegenerative diseases (NDD) are a result of changes in the chemical composition of neurons. For example, Alzheimer’s disease (AD) is the product of Aβ peptide deposition which results in changes in the ion concentration. These changes in ion concentration affect the responses of the neuron to stimuli and often result in inducing excessive excitation or inhibition. This paper investigates the dynamics of a single neuron as ion changes occur. These changes are incorporated using the Nernst equation. Within the central and peripheral nervous system, signals and hence rhythms, are propagated through the coupling of the neurons. It was found that under certain conditions the coupling strength between two neurons could mitigate changes in ion concentration. By defining the state of perfect synchrony, it was shown that the effect of ion imbalance in coupled neurons was reduced while in uncoupled neurons these changes had a more significant impact on the neuronal behavior.
IFAC Proceedings Volumes, 1999
Proceedings of the International Joint Conference on Neural Networks, 2003.
Cluster Analysis is traditionally an unsupervised data reduction technique. However, hy unifying ... more Cluster Analysis is traditionally an unsupervised data reduction technique. However, hy unifying ideas from both Cluster Analysis and Rough Set Theory, the inherent structure of a data set can be expressed as a rule set which, in turn, can be modified in a supervised manner to obtain a decision .rule set with minimal ambiguity. This process of encasing knowledge extraction in an algorithmic framework results in an optimal process for decision support.
International Conference on Control '94, 1994
ABSTRACT Three types of recurrent network configurations have been proposed since they enable ade... more ABSTRACT Three types of recurrent network configurations have been proposed since they enable adequate description of temporal behaviour. The concept of relative order has been introduced so as to provide a framework for analysing such network configurations. It has been demonstrated that, excluding pathologies, each configuration is of relative order unity. Therefore it follows that an inverse exists for each type of configuration (Tsinias and Kalouptsidis, 1983; Hirschorn, 1979).
Biological Cybernetics, 2003
Fifth International Conference on FACTORY 2000 - The Technology Exploitation Process, 1997
The modelling and control of systems with linear mathematics is a well established field. However... more The modelling and control of systems with linear mathematics is a well established field. However, for applications in the real world, the development of nonlinear modelling and control techniques continues. Following the successful integration of a nonlinear modelling capability, using neural networks, into the Connoisseur/sup TM/ advanced control package, this paper describes the development and integration of a genetic algorithm (GA) based nonlinear controller. The control problem is outlined, an overview of genetic algorithms is given, and the use of a genetic algorithm as a controller is explained. Example results are shown, with comments about the various factors that affect the performance of the controller, and the practical implementation of the controller.
HAL (Le Centre pour la Communication Scientifique Directe), 2006
IFAC Proceedings Volumes, 2006
The last decade has seen a rapid increase and convergence in the use of (a) digital network techn... more The last decade has seen a rapid increase and convergence in the use of (a) digital network technology, (b) embedded systems and (c) fault analysis and on-line diagnosis. This has resulted in an increase in complexity of the resultant system which must not only ensure performance but also monitor and compensate for faults and hence avoid total failure. The advances made have facilitated a "plug-and-play" (PnP) characteristic of networked embedded systems. This paper provides a framework for the analysis and development of strategies which enable a fault-tolerant PnP feature.
This paper describes the development of a genetic algorithm based nonlinear controller. It builds... more This paper describes the development of a genetic algorithm based nonlinear controller. It builds on the successful integration of the modelling capability of an artificial neural network approach within the advanced control package Connoisseur/sup TM/. The case for the long standing need for a generalised nonlinear controller for handling practical nonlinear and time varying systems is made. The motivation in terms of achieving tighter control, leading to increased efficiency and profitability are stressed. Existing techniques of gain scheduling and multiple models are briefly discussed, as well as their limitations. In the approach adopted in this paper the genetic algorithm based controller is employed to search for an optimal set of control outputs to minimise a given performance index. The paper gives examples of simulation studies and comments of the various factors that affect the performance of the controller and the practical implementation of the controller.
AI, Computer Science and Robotics Technology
This article provides an optimisation method using a Genetic Algorithm approach to apply feature ... more This article provides an optimisation method using a Genetic Algorithm approach to apply feature selection techniques for large data sets to improve accuracy. This is achieved through improved classification, a reduced number of features, and furthermore it aids in interpreting the model. A clinical dataset, based on heart failure, is used to illustrate the nature of the problem and to show the effectiveness of the techniques developed. Clinical datasets are sometimes characterised as having many variables. For instance, blood biochemistry data has more than 60 variables that have led to complexities in developing predictions of outcomes using machine-learning and other algorithms. Hence, techniques to make them more tractable are required. Genetic Algorithms can provide an efficient and low numerically complex method for effectively selecting features. In this paper, a way to estimate the number of required variables is presented, and a genetic algorithm is used in a “wrapper” form...