https://dspace.iiti.ac.in:8080/jspui/handle/123456789/3641 2026-05-12T17:10:18Z https://dspace.iiti.ac.in:8080/jspui/handle/123456789/18027 Title: FUMESNet: Exploring Frequency-Based Transformer and Improving Skip Connection for Hyperspectral Methane Plume Segmentation Authors: Dixit, Aditya; Gupta, Puneet Abstract: Accurate segmentation of methane (CH4) plumes from hyperspectral imagery (HSI) plays an important role in emission monitoring and environmental assessment. CH4 plume signatures appear as subtle spatial intensity variations accompanied by characteristic spectral responses across specific wavelength ranges, making their delineation challenging under complex background and atmospheric conditions. Spatial-frequency analysis can enhance structural and textural representations of plume regions, helping to suppress background clutter and preserve fine boundary details. However, many existing segmentation approaches do not explicitly exploit spatial-frequency information within the attention mechanism, and encoder-decoder architectures with skip connections often propagate redundant features due to high interband correlation, degrading plume-background separation. To address these limitations, we propose a frequency-based Transformer and improving skip connection for hyperspectral methane plume segmentation (FUMESNet), a Transformer-based framework that jointly leverages spatial context, spatial-frequency representations, and channel-wise feature relevance for CH4 plume segmentation. Specifically, an adaptive spatial-frequency integration (ASFI) module incorporates magnitude and phase information derived from spatial-frequency analysis into the attention computation, improving the delineation of weak and diffuse plume structures while preserving boundary integrity. In addition, an efficient global channel attention (EGCA) module is embedded within skip connections to adaptively emphasize spectrally informative and spatially relevant channels, mitigating the influence of redundant features during feature fusion. Experiments on multiple HSI datasets demonstrate that FUMESNet achieves consistent performance improvements over state-of-the-art (SOTA) methods, and ablation studies further confirm the complementary contributions of ASFI and EGCA to CH4 plume segmentation. © 1963-2012 IEEE. 2026-01-01T00:00:00Z https://dspace.iiti.ac.in:8080/jspui/handle/123456789/18009 Title: Malware Detection Using Hybrid Vision Transformer with CNN backbone Authors: Sharmila, S. P.; Chaudhari, Narendra Shivaji Abstract: Traditional signature-based malware detection techniques struggle to keep up with the evolving nature of malicious software. These traditional methods often fail to detect novel, obfuscated, or polymorphic malware, posing significant challenges for cybersecurity professionals. As a result, there has been a growing shift toward intelligent, learning-based techniques that can generalize across malware variants. In this paper, we propose a novel hybrid deep learning model that leverages the power of Convolutional Neural Networks (CNNs) and Vision Transformers (ViT) to enhance malware detection. The approach begins by converting executable (.exe) files into grayscale images, thereby transforming raw binary data into a visual domain compatible with modern computer vision models. CNNs are utilized for local feature extraction, capturing spatial relationships such as byte-level patterns and textures within the image, while ViT models global dependencies and contextual relationships across the image using self-attention mechanisms. This synergy enables the model to effectively detect both simple and complex malware families with high accuracy. The proposed architecture is evaluated on a benchmark malware image dataset and demonstrates superior performance, achieving an accuracy of 98.22% and precision 98.35%, along with excellent F1-Score and recall capability. The results indicate the model's robustness, scalability, and potential for integration into real-time malware detection systems to enhance cybersecurity infrastructure. © 2025 IEEE. 2025-01-01T00:00:00Z https://dspace.iiti.ac.in:8080/jspui/handle/123456789/17994 Title: Brief Announcement: Optimal Dispersion Under Asynchrony Authors: Pattanayak, Debasish Abstract: We study the dispersion problem in anonymous port-labeled graphs: k ≤ n mobile agents, each with a unique ID and initially located arbitrarily on the nodes of an n-node graph with maximum degree Δ, must autonomously relocate so that no node hosts more than one agent. Dispersion serves as a fundamental task in the distributed computing of mobile agents, and its complexity stems from key challenges in local coordination under anonymity and limited memory. The goal is to minimize both the time to achieve dispersion and the memory required per agent. It is known that any algorithm requires Ω(k) time in the worst case, and Ω(log k) bits of memory per agent. A recent result [9] gives an optimal O(k)-time algorithm in the synchronous setting and an O(k log k)-time algorithm in the asynchronous setting, both using O(log(k + Δ)) bits. We close the complexity gap in the asynchronous setting by presenting the first dispersion algorithm that runs in optimal O(k) time using O(log(k + Δ)) bits of memory per agent. Our solution relies on a novel technique for constructing a port-one tree in anonymous graphs, which may be of independent interest. © 2025 Debasish Pattanayak, Ajay D. Kshemkalyani, Manish Kumar, Anisur Rahaman Molla, and Gokarna Sharma. 2025-01-01T00:00:00Z https://dspace.iiti.ac.in:8080/jspui/handle/123456789/17968 Title: Transformer-aware sequence-to-sequence network for personalized tag recommendation in software information sites Authors: Bansal, Shubhi; Sunchu, Jahnavi; Dar, Shahid Shafi; Kumar, Nagendra Abstract: Context: Automatic tag recommendation is crucial for content understanding and retrieval on software information sites. Existing approaches formulate tag recommendation as either multi-label classification problem or sentence matching. However, multi-label classification by treating tags as independent labels, neglects semantic relationships and dependencies among them, leading to inconsistent recommendations. Sentence matching techniques, which rely on lexical similarity, fail to capture contextual information and broader semantic meaning. Moreover, several works leverage limited content sources such as title, body, and code snippet of software objects, leading to data sparsity issues. Extant research provides generic tag suggestions, overlooking users’ expertise and interests. Objective: To address these limitations, we propose a novel trnsformer-based sequence-to-sequence framework for personlaised tag recommendation, dubbed as ANNOTATION. Methods: This approach enables the model to learn dependencies between tags and generate contextually relevant recommendations. To enhance the representation of software objects and mitigate data sparsity, we incorporate valuable information from associated comments, such as clarifications, usage examples, and bug reports. This additional conversational context provides insights into the problem, alternative solutions, and related concepts discussed by the community, resulting in more informed tag recommendations. Furthermore, we personalize tag suggestions by incorporating user profile descriptions and badges, which reflect users’ expertise and interests within specific domains. This ensures that generated tags align with both the software object and the user’s specific knowledge domain, contributing to a more tailored user experience. Results: Extensive empirical and qualitative evaluations on datasets from Code Review and Stack Overflow demonstrate that our approach significantly outperforms state-of-the-art methods. Conclusion: Our findings highlight the importance of considering tag dependencies, contextual information, and user preferences for accurate and personalized tag recommendation in software information sites. © 2026 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies. 2026-01-01T00:00:00Z