Browsing by Author "Annappa, B."

Now showing 1 - 20 of 168

3D AttU-NET for Brain Tumor Segmentation with a Novel Loss Function
(Institute of Electrical and Electronics Engineers Inc., 2023) Roy, R.; Annappa, B.; Dodia, S.
In the United States of America (USA), every year 150,000 patients are registered with a secondary brain tumor that is not generated in the brain. This necessitates the need for early brain tumor detection, which in turn will help patients to live longer. For clinical evaluation and treatment, precise segmentation of brain tumors in MRI images is required. This process can be aided by machine learning and efficient image processing, but manual imaging can be time-consuming. In this study, we aim to develop an 3D automated segmentation algorithm with a novel loss function. A 3D attention UNET CNN model was trained using the novel loss function, which was calculated by taking the weighted average of dice loss and focal loss to overcome the class imbalance. Results show the enhancement in the segmentation performance of attention UNET model with an average increase of 5% in the Dice coefficient for all three classes. However, the model's performance was not as strong for enhanced and core tumors. Further research may be needed to optimize performance in these areas. . Â© 2023 IEEE.
A comprehensive review of facial expression recognition techniques
(Springer Science and Business Media Deutschland GmbH, 2023) Rashmi Adyapady, R.R.; Annappa, B.
Emotion recognition has opened up many challenges, which lead to various advances in computer vision and artificial intelligence. The rapid development in this field has encouraged the development of an automatic system that could accurately analyze and measure the emotions of human beings via facial expressions. This study mainly focuses on facial expression recognition from visual cues, as visual information is the most prominent channel for social communication. The paper provides a comprehensive review of recent advancements in algorithm development, presents the overall findings performed over the past decades, discusses their advantages and constraints. It explores the transition from the laboratory-controlled environment to challenging real-world (in-the-wild) conditions, focusing on essential issues that require further exploration. Finally, relevant opportunities in this field, challenges, and future directions mentioned in this paper assist the researchers and academicians in designing efficient and robust facial expression recognition systems. © 2022, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.
A Dual Phase Approach for Addressing Class Imbalance in Land-Use and Land-Cover Mapping From Remotely Sensed Images
(Institute of Electrical and Electronics Engineers Inc., 2024) Putty, A.; Annappa, B.; Prajwal, R.; Pariserum Perumal, S.P.
Semantic segmentation of remotely sensed images for land-use and land-cover classes plays a significant role in various ecosystem management applications. State-of-the-art results in assigning land-use and land-cover classes are primarily achieved using fully convolutional encoder-decoder architectures. However, the uneven distribution of the land-use and land-cover classes becomes a major hurdle leading to performance skewness towards majority classes over minority classes. This paper proposes a novel dual-phase training, with the first phase proposing a new undersampling technique using minority class focused class normalization and the second phase that uses this learnt knowledge for ensembling to prevent overfitting and compensate for the loss of information due to undersampling. The proposed method achieved an overall performance gain of up to 2% in MIoU, Kappa, and F1 Score metrics and up to 3% in class-wise F1-score when compared to the baseline models on Wuhan Dense Labeling, Vaihingen and Potsdam datasets. © 2013 IEEE.
A fuzzy sectional real-time scheduling algorithm based on system load
(Springer Verlag service@springer.de, 2013) Annappa, B.
Earliest Deadline First (EDF) Algorithm is one of the most widely known dynamic real-time task scheduling algorithms. However, when a real-time system is overloaded, experiments and analysis have proved that EDF algorithm is ineffective. Considering the algorithm's instability during the practical task executing environment in an overloaded state, it is necessary to apply a few decision making techniques to ensure a good overall performance. In this paper, we propose a dynamic sectional real-time scheduling algorithm called Fuzzy Sectional Scheduling (FSS), which identifies the system load and employs suitable scheduling techniques to improve overall performance. The simulation results show that the Fuzzy Sectional Scheduling Algorithm could improve the real-time system performance to a considerably greater extent compared to the classical algorithms such as EDF, HVF (Highest Value First) and HDF (Highest Density First) algorithms; under all workload conditions. Â© 2013 Springer.
A Healthcare management using clinical decision support system
(Institute of Electrical and Electronics Engineers Inc., 2018) Thomas, L.; Manoj Kumar, M.V.; Annappa, B.
From the literature it is studied that, most of the medical error is due to faulty healthcare system. Due to this, there is treatment delay, that leads to complications in later stages of disease progression. Medical error caused due to the failure in healthcare system can be reduced by employing an appropriate clinical decision support system (CDSS). CDSS helps in identifying the severity of disease by predicting its progression. The treatment management of gallstone disease is considered as a case study in this paper.This paper presents a CDSS with the help of machine learning for improving the treatment management. CDSS with the help of a statistical comparator, identifies an efficient tool for finding the associated risk factors. These risk factors are then used to predict the disease progression and identify the cases that may need Endoscopic Retrograde Cholangio-Pancreatography (ERCP) as the treatment progresses. The model that learns and predicts accurately is selected, using the concept of Area Under Curve (AUC). For this purpose, a Modified Cascade Neural Network (ModCNN) built upon the architecture of Cascade-Correlation Neural Network (CCNN) is proposed and tested using an ADAptive LInear NEuron (ADALINE) circuit. It's performance is evaluated and compared with Artificial Neural Network (ANN) and CCNN.Using this prediction information, disease progression is analysed and proper treatment is initiated, thereby reducing the medical error. ModCNN showed better accuracy (96.42%) for predicting the disease progression when compared with CCNN (93.24%) and ANN (89.65%). Thus, CDSS presented here, assisted in reducing the medical error and providing better healthcare management. Â© 2018 IEEE.
A holistic approach to influence maximization
(Springer International Publishing, 2017) Sumith, N.; Annappa, B.; Bhattacharya, S.
A social network is an Internet-based collaboration platform that plays a vital role in information spread, opinion-forming, trend-setting, and keeps everyone connected. Moreover, the popularity of web and social networks has interesting applications including viral marketing, recommendation systems, poll analysis, etc. In these applications, user influence plays an important role. This chapter discusses how effectively social networks can be used for information propagation in the context of viral marketing. Picking the right group of users, hoping they will cause a chain effect of marketing, is the core of viral marketing applications. The strategy used to select the correct group of users is the influence maximization problem. This chapter proposes one of the viable solutions to influence maximization. The focus is to find those users in the social networks who would adopt and propagate information, thus resulting in an effective marketing strategy. The three main components that would help in the effective spread of information in the social networks are: the network structure, the user's influence on others, and the seeding algorithm. Amalgamation of these three aspects provides a holistic solution to influence maximization. © Springer International Publishing AG 2017. All rights reserved.
A holistic approach to influence maximization in social networks: STORIE
(Elsevier Ltd, 2018) Sumith, N.; Annappa, B.; Bhattacharya, S.
Crowd sourcing techniques are used in social networks to propagate information at a faster pace through campaigns. One of the challenges of crowd sourcing system is to recruit right users to be a part of successful campaigns. Fetching this right group of people, who influence a vast population to adopt information, is termed as influence maximization. Concerns of scalability and effectiveness need an effective and a viable solution. This paper proposes the solution in three stages. At the first stage, the large social network is pruned based on the nodal properties to make the solution scalable. At the second stage, Outdegree Rank (OR), is proposed and at the third stage, Influence Estimation (IE) approach estimates user influence. This work amalgamates aspects of structure, heuristic and user influence, to form STORIE. The proposed approach is compared to standard heuristics, on various experimental setups such as RNNDp, RNUDp and TVM. The spread of information is observed for HEP, PHY, Twitter, Infectious and YouTube data, under Independent Cascade model and STORIE gives optimal results, with an increase up to 50%. Although the paper discusses influence maximization, the proposed approach is also applicable to understand the spread of epidemics, computer virus, and rumor spreading in the real world and can also be extended to detect anomalies in web and social networks. © 2017 Elsevier B.V.
A Multimodal Contrastive Federated Learning for Digital Healthcare
(Springer, 2023) Sachin, D.N.; Annappa, B.; Ambesenge, S.; Tony, A.E.
Digital healthcare applications have gained enormous global interest due to the rapid development of the internet of medical things (IoMT), which helps access massive amounts of multimodal healthcare data. Using this rich multimodal data without violating user privacy becomes crucial. Federated learning (FL) isolates data and protects user privacy. Clients collaboratively learn global models without data transmission. Most of the current FL approaches still depend on single-modal data. It is known that multimodal data always benefit from the complementarity of different modalities. This paper proposes a multimodal contrastive federated learning framework for digital healthcare. The proposed framework solves the multimodal federated learning problem. The proposed architecture used a geometric multimodal contrastive representation learning method to learn representations of multiple modalities in a shared, high-dimensional space. This helps optimize the representations to capture the inter-modal relationships better and improves the multimodal model’s overall performance. Experiments show that the proposed framework performs better than conventional single-modality FL and multimodal FL framework approaches. Given its generality and extensibility, the proposed framework can be used for many downstream tasks in healthcare applications. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.
A novel receptive field-regularized V-net and nodule classification network for lung nodule detection
(John Wiley and Sons Inc, 2022) Dodia, S.; Annappa, B.; Mahesh, M.
Recent advancements in deep learning have achieved great success in building a reliable computer-aided diagnosis (CAD) system. In this work, a novel deep-learning architecture, named receptive field regularized V-net (RFR V-Net), is proposed for detecting lung cancer nodules with reduced false positives (FP). The method uses a receptive regularization on the encoder block's convolution and deconvolution layer of the decoder block in the V-Net model. Further, nodule classification is performed using a new combination of SqueezeNet and ResNet, named nodule classification network (NCNet). Postprocessing image enhancement is performed on the 2D slice by increasing the image's intensity by adding pseudo-color or fluorescence contrast. The proposed RFR V-Net resulted in dice similarity coefficient of 95.01% and intersection over union of 0.83, respectively. The proposed NCNet achieved the sensitivity of 98.38% and FPs/Scan of 2.3 for 3D representations. The proposed NCNet resulted in considerable improvements over existing CAD systems. © 2021 Wiley Periodicals LLC.
A scalable cloud platform using matlab distributed computing server integrated with HDFS
(IEEE Computer Society help@computer.org, 2012) Dutta, R.; Annappa, B.
The Hadoop Distributed File System (HDFS) is a large data storage system which exhibits several features of a good distributed file system. In this paper we integrate Matlab Distributed Computing Server (MDCS) with HDFS to build a scalable, efficient platform for scientific computations. We use an FTP server on top of HDFS for data transfer from the Matlab system to HDFS. The motivation of using HDFS for storage with MDCS is to provide an efficient, fault-tolerant file system and also to utilize the resources efficiently by making each system serve as both data node for HDFS and worker for MDCS. We test the storage efficiency of HDFS and compare with normal file system for data transfer operations through MDCS. Â© 2012 IEEE.
A secure and scalable framework for group communication
(2006) Annappa, B.; Rani, G.P.
Critical networking issues in group communication involve security, scalability and dynamic membership changes. Security provides confidentiality, authenticity, and integrity of messages exchanged between group members. Tree Group Diffie Hellman is an efficient key agreement protocol for peer groups. But, it is not scalable beyond 100 members. On the other hand, large multicast groups don't support dynamic membership changes. In this Paper, we propose a new framework, which addresses the problems of scalability and dynamic membership change. Â© 2006 IEEE.
A Workflow Scheduling Approach With Modified Fuzzy Adaptive Genetic Algorithm in IaaS Clouds
(Institute of Electrical and Electronics Engineers Inc., 2023) Rizvi, N.; Ramesh, D.; Wang, L.; Annappa, B.
The emergence of the cloud platform with substantial resources to offer on-demand instigated the researchers to migrate the scientific workflows to the cloud environment. The scheduling of workflows with diverse QoS parameters is not a trivial task, but an NP-Complete problem. Several heuristics for QoS constrained workflows have been investigated. However, most of them focus only on time and cost and do not guarantee high resource utilization. The scheduling of the workflow tasks over the minimum cloud resources under the defined time limit is a grave concern. In this article, an algorithm named MFGA (Modified Fuzzy Adaptive Genetic Algorithm) has been formulated to minimize the makespan and improve resource utilization under both deadline and budget constraints. A fuzzy logic controller has also been devised to control the crossover and mutation rates that prevent MFGA from getting stuck in a local optimum. MFGA has a novel crossover technique that adds the fittest solutions in the population. Additionally, a new mutation technique has also been introduced, which minimizes the makespan and increases the reusability of the resources. The simulation experiments with the real workflows show that the proposed MFGA outperforms other state-of-the-art algorithms. © 2008-2012 IEEE.
Abdominal Multi-Organ Segmentation Using Federated Learning
(Institute of Electrical and Electronics Engineers Inc., 2024) Yadav, G.; Annappa, B.; Sachin, D.N.
Multi-organ segmentation refers to precisely de-lineating and identifying multiple organs or structures within medical images, such as Computed Tomography (CT) scans or Magnetic Resonance Imaging (MRI), to outline boundaries and regions for each organ accurately. Medical imaging is crucial to comprehending and diagnosing a wide range of illnesses for which accurate multi-organ image segmentation is often required for successful analysis. Due to the delicate nature of medical data, traditional methods for multi-organ segmentation include centralizing data, which presents serious privacy problems. This centralized training strategy impedes innovation and collaborative efforts in healthcare by raising worries about patient confidentiality, data security, and reg-ulatory compliance. The development of deep learning-based image segmentation algorithms has been hindered by the lack of fully annotated datasets, and this issue is exacerbated in multi-organ segmentation. Federated Learning (FL) addresses privacy concerns in multi-organ segmentation by enabling model training across decentralized institutions without sharing raw data. Our proposed FL-based model for CT scans ensures data privacy while achieving accurate multi-organ segmentation. By leveraging FL techniques, this paper collaboratively trains segmentation models on local datasets held by distinct medical institutions. The expected outcomes encompass achieving high Dice Similarity Coefficient (DSC) metrics and validating the efficacy of the proposed FL approach in attaining precise and accurate segmentation across diverse medical imaging datasets. Â© 2024 IEEE.
ACR2UNet: Semantic Segmentation of Remotely Sensed Images using Residual-Recurrent UNet and Asymmetric Convolutions
(Institute of Electrical and Electronics Engineers Inc., 2023) Putty, A.; Annappa, B.
Land-use and land-cover (LULC) mapping is one of the significant components in environmental monitoring. LULC mapping, necessary to manage the vital resource of land, has been achieved, in recent years, by segmenting remotely sensed images (RSIs). A standard paradigm for segmentation is UNet, and this paper proposes a novel asymmetric convolutional residualrecurrent UNet architecture, which utilizes the power of asymmetric convolutions as well as residual and recurrent techniques for mapping RSIs. The proposed methodology has a couple of additional advantages. First, asymmetric convolution operations strengthen the square kernels and enhance the semantic feature space. Further, a recurrent network assists in providing rich local contextual information with the help of residual inputs. The presented model is evaluated on the WHDLD dataset for LULC segmentation and is found to achieve an improvement of 1-2% in the mIoU score compared to state-of-the-art methods. Â© 2023 IEEE.
Adaptive GPU resource scheduling on virtualized servers in cloud gaming
(2018) Yadav, H.; Annappa, B.
Cloud Gaming is a modern approach for providing high-quality gaming service to the user. Cloud Gaming renders the sophisticated game scene on powerful servers and streams the video in real-time to the thin clients. Cloud Based on-demand gaming provides an affordable and flexible solution to the users with limited computing resources and enables the gamer to play high-end graphic games on low-end thin clients without upgrading their existing system. Due to its advantages, it has attracted both researchers in academia and industry. Virtualization undertakes a key part in cloud gaming providing multiple users and application to share the resource. Yet the Graphical processing unit, on which these high-end games rely, is difficult to virtualize because of its sophisticated architecture design and different device drivers. Application of GPU virtualization is very limited in data centers due to its inefficient scheduling mechanism, i.e. First come First Serve algorithm. Due to this, the virtual machines are not able to satisfy their Service level agreement. Existing cloud gaming data centers rely on a non-virtualized environment with dedicated GPU and sophisticated hardware. In this paper, an adaptive scheduling mechanism of the virtual machine for cloud gaming is proposed to achieve efficient GPU resource management in data centers, hosting cloud gaming servers. � 2017 IEEE.
Adaptive GPU resource scheduling on virtualized servers in cloud gaming
(Institute of Electrical and Electronics Engineers Inc., 2017) Yadav, H.; Annappa, B.
Cloud Gaming is a modern approach for providing high-quality gaming service to the user. Cloud Gaming renders the sophisticated game scene on powerful servers and streams the video in real-time to the thin clients. Cloud Based on-demand gaming provides an affordable and flexible solution to the users with limited computing resources and enables the gamer to play high-end graphic games on low-end thin clients without upgrading their existing system. Due to its advantages, it has attracted both researchers in academia and industry. Virtualization undertakes a key part in cloud gaming providing multiple users and application to share the resource. Yet the Graphical processing unit, on which these high-end games rely, is difficult to virtualize because of its sophisticated architecture design and different device drivers. Application of GPU virtualization is very limited in data centers due to its inefficient scheduling mechanism, i.e. First come First Serve algorithm. Due to this, the virtual machines are not able to satisfy their Service level agreement. Existing cloud gaming data centers rely on a non-virtualized environment with dedicated GPU and sophisticated hardware. In this paper, an adaptive scheduling mechanism of the virtual machine for cloud gaming is proposed to achieve efficient GPU resource management in data centers, hosting cloud gaming servers. Â© 2017 IEEE.
Algorithmic approach for strategic cell tower placement
(IEEE Computer Society help@computer.org, 2015) Kashyap, R.; Bhuvan, B.M.; Chamarti, S.; Bhat, T.; Jothish, M.; Annappa, B.
The increasing number of cell phone users and the usage of cell phones in remote areas have demanded the network service providers to increase their coverage and extend it to all places. Cost of placing a cell tower depends on the height and location, and as it can be very expensive, they have to be placed strategically to minimize the cost. The research aims to find a simple implementable algorithm which effectively determines the strategic positions of the cell towers. Given a satellite image and population density, and obtaining topographical information from GIS (Geographic Information Systems), potential tower locations can be determined. Applying the proposed three stage algorithm, out of many potential tower locations only the indispensible and optimal locations can be chosen. In addition, this algorithm helps to find out the optimal height of the tower at a chosen potential tower location. Hence, the proposal will provide cost-effective way for tower placement specifying their optimal position and height to cover any area and population. Â© 2014 IEEE.
Alphabetic cryptography: Securing communication over cloud platform
(2019) Cowlessur, S.K.; Annappa, B.; Manoj, Kumar, M.V.; Thomas, L.; Sneha, M.M.; Puneetha, B.H.
This paper introduces alphabetic cryptography inspired by bidirectional DNA encryption algorithm. Alphabetic cryptography first offers higher randomization and secure communication over the cloud computing platform, and second supports the exchange of complete UNICODE character set. Alphabetic cryptography has been implemented on mobile and desktop platforms. Through experimental studies, it has been observed that randomness of encryption increases exponentially with the increase in the number of alphabets of the alphabetic encryption scheme. � Springer Nature Singapore Pte Ltd. 2019.
Alphabetic cryptography: Securing communication over cloud platform
(Springer Verlag service@springer.de, 2019) Cowlessur, S.K.; Annappa, B.; Manoj Kumar, M.V.; Thomas, L.; Sneha, M.M.; Puneetha, B.H.
This paper introduces alphabetic cryptography inspired by bidirectional DNA encryption algorithm. Alphabetic cryptography first offers higher randomization and secure communication over the cloud computing platform, and second supports the exchange of complete UNICODE character set. Alphabetic cryptography has been implemented on mobile and desktop platforms. Through experimental studies, it has been observed that randomness of encryption increases exponentially with the increase in the number of alphabets of the alphabetic encryption scheme. Â© Springer Nature Singapore Pte Ltd. 2019.
An Efficient Deep Transfer Learning Approach for Classification of Skin Cancer Images
(Springer Science and Business Media Deutschland GmbH, 2023) Naik, P.P.; Annappa, B.; Dodia, S.
Prolonged exposure to the sun for an extended period can likely cause skin cancer, which is an abnormal proliferation of skin cells. The early detection of this illness necessitates the classification of der-matoscopic images, making it an enticing study problem. Deep learning is playing a crucial role in efficient dermoscopic analysis. Modified version of MobileNetV2 is proposed for the classification of skin cancer images in seven classes. The proposed deep learning model employs transfer learning and various data augmentation techniques to more accurately classify skin lesions compared to existing models. To improve the perÂ¬formance of the classifier, data augmentation techniques are performed on â€œHAM10000" (Human Against Machine) dataset to classify seven difÂ¬ferent kinds of skin cancer. The proposed model obtained a training accuracy of 96.56% and testing accuracy of 93.11%. Also, it has a lower number of parameters in comparison to existing methods. The aim of the study is to aid dermatologists in the clinic to make more accurate diagnoses of skin lesions and in the early detection of skin cancer. Â© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.