Papers by dheevatsa mudigere
We argue that the current heterogeneous computing environment mimics a complex nonlinear system
w... more We argue that the current heterogeneous computing environment mimics a complex nonlinear system
which needs to borrow the concept of time-scale separation and the delayed difference approach from
statistical mechanics and nonlinear dynamics. We show that by replacing the usual difference equations
approach by a delayed difference equations approach, the sequential fraction of many scientific computing
algorithms can be substantially reduced. We also provide a comprehensive theoretical analysis to establish
that the error and stability of our scheme is of the same order as existing schemes for a large, well-
characterized class of problems.
Nature Inspired Techniques for Identification of Helicopter Dynamics Based on Flight Data
ABSTRACT The complexity of helicopter flight dynamics makes modeling and helicopter system identi... more ABSTRACT The complexity of helicopter flight dynamics makes modeling and helicopter system identification a very difficult task. Most of the traditional techniques used for nonlinear dynamical system identification require a model structure to be defined a priori. In case of helicopter dynamics, defining a priori model is difficult due to its complexity and the interplay between various subsystems. To overcome this difficulty, non-parametric approaches are commonly adopted for helicopter system identification. Artificial Neural Network are a widely used class of algorithms for non-parametric system identification, among them, the Nonlinear Auto Regressive eXogeneous input network (NARX) model is very popular because of its proven stability characteristics. But, this model also necessitates some in-depth knowledge regarding the system being modelled. This forms one of the major drawbacks of this very popular model. There have been many approaches proposed to circumvent this drawback, and still try to retain its advantageous characteristics. In this paper we carry out an extensive study of one such newly proposed approach for circumventing this problem in the NARX model. Using a modified NARX model with a II-Tiered, externally driven recurrent neural network architecture. This model uses the NARX model as the substrate; single hidden layer perceptron network based architecture along with tapped delay lines is employed for modelling the dynamical systems. This is coupled with an outer optimization routine for evolving the order of the system. In this paper this generic architecture is comprehensively explored to ascertain its usability and critically asses it’s potential. Different implementations of this architecture, based on nature inspired computational techniques (– Swarm theory, Artificial Immune systems and Artificial Bee colony) are proposed in this paper and have been critically compared and evaluated. All the proposed models are capable of identifying the helicopter dynamics without any a priori knowledge about the system under study. Thus, successfully circumventing one of the major drawbacks of the very popular NARX model, without losing its numerous advantages. Simulations have been carried out for identifying the longitudinally uncoupled dynamics. Results of identification indicate a quite close correlation between the actual and the predicted response of the helicopter for all the models. Further, these different models are compared with each other and with the conventional NARX model, to assess them and critically appraise them with respect to accuracy, stability, robustness, efficiency and other such issues.
Quantification of the effect of therapeutic yoga by postural analysis using Image
Nature inspired optimization techniques for the design optimization of laminated composite structures using failure criteria
Expert Systems with …, Jan 1, 2011
... for the Design optimization of Laminated Composite Structures using failure Criteria G.Naraya... more ... for the Design optimization of Laminated Composite Structures using failure Criteria G.Narayana Naik , SNOmkar, Dheevatsa Mudigere ... Therefore, in this work, the Failure Mechanism Based Failure Criterion (FMBFC) developed by the authors (Krishna Murty, Narayana Naik, & ...
Proceedings of the International …, Jan 1, 2007
In this paper we present a generic model for dynamical system identification based on the Particl... more In this paper we present a generic model for dynamical system identification based on the Particle Swarm Optimization (PSO) algorithm. This model is capable of identifying dynamical systems given just a set of input-output values corresponding to the system and does not require any a priori knowledge of the system under study. In this model, single hidden layer perceptron network based architecture along with tapped delay lines is used for modelling the dynamical systems. This is coupled with an additional PSO algorithm for evolving the order of the system under study. The PSO algorithm with its stochastic means identifies the order, thus determining the configuration of the network used to model the system. Further, a PSO based learning algorithm is used to train the network. This model successfully circumvents one the major drawbacks of the very popular Nonlinear Auto Regressive eXogeneous input network (NARX) model for dynamical system identification, at the same time retaining its numerous advantages. Also, the use of evolutionary algorithms such as PSO renders the model robust and flexible.
Crop classifieation using bj010 eallyinspired techniques with hi resolution satelliteimage
Journal oftheIndian SocietyofRemote …
Journal of Bodywork and …, Jan 1, 2007
A generic methodology to compare and quantify two photographs of body posture, taken under non-st... more A generic methodology to compare and quantify two photographs of body posture, taken under non-standard conditions is proposed in the current work. Suitable digital image processing tools are employed to accentuate the use of imaging technology to assess the posture. ImageJ, a versatile public domain imageprocessing tool has been employed to carry out the postural analysis. In this paper, we have described two case studies. Postural analysis is carried out on the portrait and profile photographs of the subjects taken arbitrarily, before and after one yoga session to demonstrate the method. Since this study deals with the photographs of the subject, which are arbitrarily taken, emphasis is laid on the normalization aspect of the photographs, after which the postural assessment aspect is discussed.
Arxiv preprint arXiv: …, Jan 1, 2010
Many high performance computing algorithms are bandwidth limited, hence the need for optimal data... more Many high performance computing algorithms are bandwidth limited, hence the need for optimal data rearrangement kernels as well as their easy integration into the rest of the application. In this work, we have built a CUDA library of fast kernels for a set of data rearrangement operations. In particular, we have built generic kernels for rearranging m dimensional data into n dimensions, including Permute, Reorder, Interlace/Deinterlace, etc. We have also built kernels for generic Stencil computations on a two-dimensional data using templates and functors that allow application developers to rapidly build customized high performance kernels. All the kernels built achieve or surpass best-known performance in terms of bandwidth utilization.
Vector evaluated particle swarm optimization (VEPSO) for multi-objective design optimization of composite structures
Computers & …, Jan 1, 2008
... criteria; TsaiWu [21], Maximum Stress [1] and the Failure Mechanism based criteria [22]. ...... more ... criteria; TsaiWu [21], Maximum Stress [1] and the Failure Mechanism based criteria [22]. ... method truly generic and ensures a completely optimum solution/configuration for the given application. ... particle swarm optimization (VEPSO) [23], a variant of the classical PSO for multi ...
Uploads
Papers by dheevatsa mudigere
which needs to borrow the concept of time-scale separation and the delayed difference approach from
statistical mechanics and nonlinear dynamics. We show that by replacing the usual difference equations
approach by a delayed difference equations approach, the sequential fraction of many scientific computing
algorithms can be substantially reduced. We also provide a comprehensive theoretical analysis to establish
that the error and stability of our scheme is of the same order as existing schemes for a large, well-
characterized class of problems.
which needs to borrow the concept of time-scale separation and the delayed difference approach from
statistical mechanics and nonlinear dynamics. We show that by replacing the usual difference equations
approach by a delayed difference equations approach, the sequential fraction of many scientific computing
algorithms can be substantially reduced. We also provide a comprehensive theoretical analysis to establish
that the error and stability of our scheme is of the same order as existing schemes for a large, well-
characterized class of problems.