Conference Papers

Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506

Browse

Author: search.filters.author.Keerthan Kumar, T.G.K.

Search Results

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    MatchVNE: A Stable Virtual Network Embedding Strategy Based on Matching Theory
    (Institute of Electrical and Electronics Engineers Inc., 2023) Keerthan Kumar, T.G.K.; Srivastava, A.; Satpathy, A.; Addya, S.K.; Koolagudi, S.G.
    Network virtualization (NV) can provide greater flexibility, better control, and improved quality of service (QoS) for the existing Internet architecture by enabling heterogeneous virtual network requests (VNRs) to share the substrate network (SN) resources. The efficient assignment of the SN resources catering to the demands of virtual machines (VMs) and virtual links (VLs) of the VNRs is known as virtual network embedding (VNE) and is proven to be NP-Hard. Deviating from the literature, this paper proposes a framework MatchVNE that is focused on maximizing the revenue-to-cost ratio of VNRs by considering a blend of system and topological attributes that better capture the inherent dependencies among the VMs. MatchVNE performs a stable VM embedding using the deferred acceptance algorithm (DAA). The preference of the VMs and servers are generated using a hybrid entropy, and the technique for order of preference by similarity to ideal solution (TOPSIS) based ranking strategy for VMs and servers. The attribute weights are determined using entropy, whereas the server and VM ranking are obtained via TOPSIS. The shortest path, VL-embedding, follows VM-embedding. The simulation results show that MatchVNE outperforms the baselines by achieving a 23% boost in the average revenue-to-cost-ratio and 44% improvement in the average acceptance ratio. © 2023 IEEE.
  • No Thumbnail Available
    Item
    Stratification of Depressed and Non-Depressed Texts from Social Media using LSTM and its Variants
    (Elsevier B.V., 2024) Keerthan Kumar, T.G.K.; Anoop, R.; Koolagudi, S.G.; Rao, T.; Kodipalli, A.
    This work examines the performance of various LSTM (long short-term memory) variants on social media text data. This study evaluates the performance of LSTM models with different architectures, namely, classic LSTM, Bidirectional LSTM, Stacked LSTM, gated recurrent unit (GRU), and bidirectional GRU, on a social network dataset comprising texts extracted from multiple social media platforms. We aim to identify the most effective LSTM variant of the five considered LSTM models for text analysis through a comparative study of the models' precision, recall, F1-score, and accuracy. The research findings show that the Classic LSTM and the GRU model perform better than the other models in accuracy. In contrast, the bidirectional models (Bidirectional LSTM and Bidirectional GRU) provide better precision scores than their respective primitive models. This research has significant implications for developing more efficient models for natural language processing applications. It offers beneficial insights into the implications involving the scrutiny of depression on social media platforms through text data analysis. © 2024 Elsevier B.V.. All rights reserved.
  • No Thumbnail Available
    Item
    ECG Signal Classification using Continuous Wavelet Transform Scalogram and Convolutional Neural Network
    (Institute of Electrical and Electronics Engineers Inc., 2024) Keerthan Kumar, T.G.K.; Ogare, M.K.; Koolagudi, S.G.
    Automated classification of electrocardiogram (ECG) signals is pivotal for timely and accurate diagnosis of cardiac abnormalities. In this work weintroduces a new method for classifying electrocardiogram (ECG) signals by merging signal processing and deep learning techniques. We utilize Continuous Wavelet Transform (CWT) to convert one-dimensional ECG signals into scalogram images, capturing both temporal and frequency details. By employing transfer learning, we fine-tune a pre-trained AlexNet Convolutional Neural Network (CNN) to categorize ECG signals into three types: arrhythmia, congestive heart failure, and normal sinus rhythm. We extensively compare our method with existing approaches, demonstrating its superior performance with an accuracy of 96%. The hierarchical structure of AlexNet enables the extraction of intricate features from ECG signals, surpassing other models that suffer from shallow architectures and reliance on manual feature engineering. Our approach not only improves automated ECG analysis but also holds promise for enhancing clinical diagnosis and management of cardiovascular conditions. © 2024 IEEE.