Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
5 results
Search Results
Item IIMH: Intention Identification in Multimodal Human Utterances(Association for Computing Machinery, 2023) Keerthan Kumar, T.G.; Dhakate, H.; Koolagudi, S.G.Intention identification is a challenging problem in the field of natural language processing, speech processing, and computer vision. People often use contradictory or ambiguous words in different contexts, which can sometimes be very confusing to identify the intention behind an utterance. Intention identification has many practical applications in the fields of natural language processing, sentiment analysis, social media analysis, robotics, and human-computer interaction, where valuable insights into user behavior can be achieved by identifying intention. In this work, we propose a model to determine whether an utterance made by a person is intentional or not intentional. To achieve this, we collected a multimodal dataset containing text, video, and speech from various TV shows, movies, and YouTube videos and labeled them with their corresponding intention. Feature extraction is done at both utterance and word levels to get useful information from all three modalities. We trained the baseline model using SVM to set a benchmark performance. We designed an architecture to detect the contradiction between positive spoken words with negative facial expressions or speech to identify an utterance as non-intentional. Along with the architecture, we used different approaches for classification and got the best results with the Support vector machine (SVM) classifier using RBF kernel, with an accuracy of 78.83% and proven to be better compared to the baseline approach. © 2023 ACM.Item DeepVNE: Deep Reinforcement and Graph Convolution Fusion for Virtual Network Embedding(Institute of Electrical and Electronics Engineers Inc., 2024) Keerthan Kumar, T.G.; Kb, A.; Siddheshwar, A.; Marali, A.; Kamath, A.; Koolagudi, S.G.; Addya, S.K.Network virtualization (NV) plays a crucial role in modern network management. One of the fundamental challenges in NV is allocating physical network (PN) resources to the demands of the virtual network requests (VNRs). This process is known as a virtual network embedding (VNE) and is NP-hard. Most of the existing approaches for VNE are based on heuristic, meta-heuristic, and exact strategies with limited flexibility and the risk of being stuck in local optimal solutions. In this concern, we provide a deep reinforcement learning (DRL) and graph convolution network (GCN) fusion for VNE (DeepVNE) for maximizing the revenue-to-cost ratio. The DeepVNE takes advantage of the power of actor-critic models within the DRL framework to detect network states and provide optimal solutions matched to current conditions. DeepVNE effectively captures the structural dependencies of both VNRs and PN resources by GCNs, allowing better decision-making during the embedding. Considering several features in the agents throughout the training phase improves resource utilization. The experiments show that DeepVNE outperforms the baselines by gaining a 51% acceptance ratio and 28% revenue-to-cost ratio. © 2024 IEEE.Item SanS: Classification of Sanskrit Mantras using Speech Processing(Association for Computing Machinery, 2024) Keerthan Kumar, T.G.; Udaya, S.; Koolagudi, S.G.Chandas is a classification metric used in Sanskrit Mantras. They are essentially meant to maintain a form of rhythm for each and every mantra, which gives the hymns their distinctive chanting pattern. Sanskrit hymns (mantras) have many different classifications. Often, before chanting, they are invoked as such in the order of Rishi (author of the hymn), Chanda (rhythm), Devata (god to which invoked), and Viniyoga (use of such hymn). One such example is the Gayatri hymn, which is from Gayatri Chanda. Other examples are verses of the Bhagavad Gita, which are entirely in Anushtup Chanda. Chandas is an essential component of Sanskrit Mantras and forms an integral part of it. Knowledge of Chandas is essential for the study of Sanskrit Poetry, and without knowing what a Chanda is, one cannot analyze Sanskrit hymns. Essentially, Chandas are the meters used to keep track of the mantras. Chandas formulate the rhythm of the mantra and the way that it should be chanted. Based on the number of syllables, there are seven different Chandas, and mantras are usually classified into one of these seven. Identification of Chandas is usually specified in the beginning before chanting; however, in cases where the Chandas are not specified, one may have to do it manually, which may be cumbersome, especially for Chandas with many syllables. In this work, we propose a novel approach called Classification of Sanskrit Mantras using Speech Processing Technology (SanS) to determine the Chanda of a mantra, given the audio file of a Sanskrit mantra and using a Wavenet architecture. The proposed SanS predicts the type of Chanda by counting the number of syllables, giving 81.57% accuracy compared to other works. © 2024 Copyright held by the owner/author(s).Item Syntactic Simplification of English Sentences Using Parse Trees(Association for Computing Machinery, 2024) Keerthan Kumar, T.G.; Adhith, A.; Jayadeep, N.; Gokulraj, M.; Koolagudi, S.G.In the realm of Natural Language Processing, the quest to simplify text is paramount, aiming to enhance accessibility and comprehension for all individuals. This study delves into the realm of English text simplification through a distinctive approach leveraging syntactic parse trees, known as syntactic simplification. Our methodology tackles two key aspects: the first involves streamlining compound sentences linked by coordinating conjunctions, while the second focuses on simplifying those connected by subordinating conjunctions, namely complex sentences. With our innovative approach, we achieved a notable BLEU score of 81.9, surpassing previous endeavors in the field. In an era marked by information saturation and diverse literacy levels, the need for clear and intelligible text is undeniable. This work not only sheds light on effective text simplification but also underscores its vital role in fostering universal understanding and communication. © 2024 Copyright held by the owner/author(s).Item NORD: NOde Ranking-based efficient virtual network embedding over single Domain substrate networks(Elsevier B.V., 2023) Keerthan Kumar, T.G.; Addya, S.K.; Satpathy, A.; Koolagudi, S.G.Network virtualization (NV) allows the service providers (SPs) to partition the substrate resources in the form of isolated virtual networks (VNs) comprising multiple correlated virtual machines (VMs) and virtual links (VLs), capturing the dependencies. Though NV brought about multiple benefits, such as service isolation, improved quality-of-service (QoS), secure communication, and better utilization of substrate resources, it also introduced numerous research challenges. In this regard, one of the predominant challenges is assigning resources to the virtual components, i.e., VMs and VLs, also termed virtual network embedding (VNE). VNE comprises two closely related sub-problems, (i.) VM embedding and (ii.) VL embedding, and both the problems have been demonstrated to be NP-Hard. In the context of VNE, maximizing the revenue to cost ratio remains the focal point for the SPs as it not only boosts acceptance of VNRs but also effectively utilizes the substrate resources. However, the existing literature on VNE suffers from the following pitfalls: (i.) They only consider system resources or (ii.) limited topological attributes. However, both attributes are quintessential in accurately capturing the VNRs and the substrate network dependencies, thereby augmenting the revenue to cost ratio. This paper proposes an efficient VNE strategy called, NOde Ranking-based efficient virtual network embedding over single Domain substrate networks (NORD), to maximize the revenue to cost ratio. To address the problem of VM embedding, NORD utilizes a hybrid entropy and the technique for order of preference by similarity to ideal solution (TOPSIS) based ranking strategy for VMs and servers considering both system and topological attributes that effectively capture the dependencies. Once the ranking is generated, A greedy VM embedding followed by shortest path VL embedding completes the assignment. Simulation results confirm that NORD attains a 40% and 61% increment in average acceptance and revenue-to-cost ratios compared to the baselines. © 2023 Elsevier B.V.