2. Conference Papers
Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/1/7
Browse
61 results
Search Results
Item Modified MapReduce framework for enhancing performance of graph based algorithms by fast convergence in distributed environment(2014) Singhal, H.; Ram Mohana Reddy, GuddetiThe amount of data which is produced is huge in current world and more importantly it is increasing exponentially. Traditional data storage and processing techniques are ineffective in handling such huge data [10]. Many real life applications require iterative computations in general and in particular used in most of machine learning and data mining algorithms over large datasets, such as web link structures and social network graphs. MapReduce is a software framework for easily writing applications which process large amount of data (multi-terabyte) in parallel on large clusters (thousands of nodes) of commodity hardware. However, because of batch oriented processing of MapReduce we are unable to utilize the benefits of MapReduce in iterative computations. Our proposed work is mainly focused on optimizing three factors resulting in performance improvement of iterative algorithms in MapReduce environment. In this paper, we address the key issues based on execution of tasks, the unnecessary creation of new task in each iteration and excessive shuffling of data in each iteration. Our preliminary experiments have shown promising results over the basic MapReduce framework. The comparative study with existing solutions based on MapReduce framework like HaLoop, has also shown better performance w.r.t algorithm run time and amount of data traffic over Hadoop Cluster. � 2014 IEEE.Item Resource Provisioning Framework for IoT Applications in Fog Computing Environment(2018) Rakshith, G.; Rahul, M.V.; Sanjay, G.S.; Natesha, B.V.; Ram Mohana Reddy, GuddetiThe increasing utility of ubiquitous computing and dramatic shifts in the domain of Internet of Things (IoT) have generated the need to devise methods to enable the efficient storage and retrieval of data. Fog computing is the de facto paradigm most suitable to make efficient use of the edge devices and thus shifting the impetus from a centralized cloud environment to a decentralized computing paradigm. By utilizing fog resources near to the edge of the network, we can reduce the latency and the overheads involved in the processing of the data by deploying the required services on them. In this paper, we present resource provisioning framework which provisions the resources and also manages the registered services in a dynamic topology of the fog architecture. The results demonstrate that using fog computing for deploying services reduces the total service time. � 2018 IEEE.Item Predicting an optimal sparse matrix format for SpMV computation on GPU(2014) Neelima, B.; Ram Mohana Reddy, Guddeti; Raghavendra, P.S.Many-threaded architecture based Graphics Processing Units (GPUs) are good for general purpose computations for achieving high performance. The processor has latency hiding mechanism through which it hides the memory access time in such a way that when one warp (group of 32 threads) is computing, the other warps perform memory bound access. But for memory access bound irregular applications such as Sparse Matrix Vector Multiplication (SpMV), memory access times are high and hence improving the performance of such applications on GPU is a challenging research issue. Further, optimizing SpMV time on GPU is an important task for iterative applications like jacobi and conjugate gradient. However, there is a need to consider the overheads caused while computing SpMV on GPU. Transforming the input matrix to a desired format and communicating the data from CPU to GPU are non-trivial overheads associated with SpMV computation on GPU. If the chosen format is not suitable for the given input sparse matrix then desired performance improvements cannot be achieved. Motivated by this observation, this paper proposes a method to chose an optimal sparse matrix format, focusing on the applications where CPU to GPU communication time and pre-processing time are nontrivial. The experimental results show that the predicted format by the model matches with that of the actual high performing format when total SpMV time in terms of pre-processing time, CPU to GPU communication time and SpMV computation time on GPU, is taken into account. The model predicts an optimal format for any given input sparse matrix with a very small overhead of prediction within an application. Compared to the format to achieve high performance only on GPU, our approach is more comprehensive and valuable. This paper also proposes to use a communication and pre-processing overhead optimizing sparse matrix format to be used when these overheads are non trivial. � 2014 IEEE.Item Performance analysis of Ensemble methods on Twitter sentiment analysis using NLP techniques(2015) Kanakaraj, M.; Ram Mohana Reddy, GuddetiMining opinions and analyzing sentiments from social network data help in various fields such as even prediction, analyzing overall mood of public on a particular social issue and so on. This paper involves analyzing the mood of the society on a particular news from Twitter posts. The key idea of the paper is to increase the accuracy of classification by including Natural Language Processing Techniques (NLP) especially semantics and Word Sense Disambiguation. The mined text information is subjected to Ensemble classification to analyze the sentiment. Ensemble classification involves combining the effect of various independent classifiers on a particular classification problem. Experiments conducted demonstrate that ensemble classifier outperforms traditional machine learning classifiers by 3-5%. � 2015 IEEE.Item Optimized Object Detection Technique in Video Surveillance System Using Depth Images(2020) Shahzad, Alam, M.; Ashwin, T.S.; Ram Mohana Reddy, GuddetiIn real-time�surveillance and intrusion detection, it is difficult to rely�only on RGB image-based videos as the accuracy�of detected object is low in the low light condition and if the video surveillance area is completely dark then the object will not be detected. Hence, in this paper, we propose a method which can increase the accuracy of object detection even in low light conditions. This paper also shows how the light intensity affects the probability of object detection in RGB, depth, and infrared images. The depth information is obtained from Kinect sensor and YOLO architecture is used to detect the object in real-time. We experimented the proposed method using real-time surveillance system which gave very promising results when applied on depth images which were taken in low light conditions. Further, in real-time object detection, we cannot apply object detection technique before applying any image preprocessing. So we investigated the depth image by which the accuracy of object detection can be improved without applying any image preprocessing. Experimental results demonstrated that depth image (96%) outperforms RGB image (48%) and infrared image (54%) in extreme low light conditions. � 2020, Springer Nature Singapore Pte Ltd.Item Optimal load balancing in cloud computing by efficient utilization of virtual machines(2014) Domanal, S.G.; Ram Mohana Reddy, GuddetiLoad balancing is the major concern in the cloud computing environment. Cloud comprises of many hardware and software resources and managing these will play an important role in executing a client's request. Now a day's clients from different parts of the world are demanding for the various services in a rapid rate. In this present situation the load balancing algorithms built should be very efficient in allocating the request and also ensuring the usage of the resources in an intelligent way so that underutilization of the resources will not occur in the cloud environment. In the present work, a novel VM-assign load balance algorithm is proposed which allocates the incoming requests to the all available virtual machines in an efficient manner. Further, the performance is analyzed using Cloudsim simulator and compared with existing Active-VM load balance algorithm. Simulation results demonstrate that the proposed algorithm distributes the load on all available virtual machines without under/over utilization. � 2014 IEEE.Item Memory-based load balancing algorithm in structured peer-to-peer system(2018) Raghu, G.; Sharma, N.K.; Domanal, S.G.; Ram Mohana Reddy, GuddetiThere are several load balancing techniques which are popular used in Structured Peer-to-Peer (SPTP) systems to distribute the load among the systems. Most of the protocols are concentrating on load sharing in SPTP Systems that lead to the performance degeneration in terms of processing delay and processing time due to the lack of resources utilization. The proposed work is related to the sender-initiated load balancing algorithms which are based on the memory. Further to check the performance of the proposed load balancing algorithm, the experimental results carried out in the real-time environment with different type of network topologies in distributed environment. The proposed work performed better over existing load balancing algorithm such as Earliest Completion Load Balancing (ECLB) and First Come First Serve (FCFS) in terms of processing delay and execution time. � Springer Nature Singapore Pte Ltd. 2018.Item On demand Virtual Machine allocation and migration at cloud data center using Hybrid of Cat Swarm Optimization and Genetic Algorithm(2017) Sharma, N.K.; Ram Mohana Reddy, GuddetiThis paper deals with the energy saving at the data center using energy aware Virtual Machines (VMs) allocation and migration. The multi-objective based VMs allocation using Hybrid Genetic Cat Swarm Optimization (HGACSO) algorithm saves the energy consumption as well as also reduces resource wastage. Further consolidating VMs onto the minimal number of Physical Machines (PMs) using energy efficient VMs migration, we can shut down idle PMs for enhancing the energy efficiency at a cloud data center. The experimental results show that our proposed HGACSO VM allocation and energy efficient VM migration techniques achieved the energy efficiency and minimization of resource wastage. � 2016 IEEE.Item Novel energy efficient virtual machine allocation at data center using Genetic algorithm(2015) Sharma, N.K.; Ram Mohana Reddy, GuddetiIncreased resources utilization from clients in a smart computing environment poses a greater challenge in allocating optimal energy efficient resources at the data center. Allocation of these optimal resources should be carried out in such a manner that we can save the energy of data center as well as avoiding the service level agreement (SLA) violation. This paper deals with the design of an energy efficient algorithm for optimized resources allocation at data center using combined approach of Dynamic Voltage Frequency Scaling (DVFS) and Genetic algorithm (GA). The performance of the proposed energy efficient algorithm is compared with DVFS. Experimental results demonstrate that the proposed energy efficient algorithm consumes 22.4% less energy over a specified workload with 0% SLA violation. � 2015 IEEE.Item NLP based sentiment analysis on Twitter data using ensemble classifiers(2015) Kanakaraj, M.; Ram Mohana Reddy, GuddetiMost sentiment analysis systems use bag-of-words approach for mining sentiments from the online reviews and social media data. Rather considering the whole sentence/ paragraph for analysis, the bag-of-words approach considers only individual words and their count as the feature vectors. This may mislead the classification algorithm especially when used for problems like sentiment classification. Traditional machine learning algorithms like Naive Bayes, Maximum Entropy, SVM etc. are widely used to solve the classification problems. These machine learning algorithms often suffer from biasness towards a particular class. In this paper, we propose Natural Language (NLP) based approach to enhance the sentiment classification by adding semantics in feature vectors and thereby using ensemble methods for classification. Adding semantically similar words and context-sense identities to the feature vectors will increase the accuracy of prediction. Experiments conducted demonstrate that the semantics based feature vector with ensemble classifier outperforms the traditional bag-of-words approach with single machine learning classifier by 3-5%. � 2015 IEEE.