Showing 1–4 of 4 results for author: Vemula, S

keepitsimple at SemEval-2025 Task 3: LLM-Uncertainty based Approach for Multilingual Hallucination Span Detection

Authors: Saketh Reddy Vemula, Parameswari Krishnamurthy

Abstract: Identification of hallucination spans in black-box language model generated text is essential for applications in the real world. A recent attempt at this direction is SemEval-2025 Task 3, Mu-SHROOM-a Multilingual Shared Task on Hallucinations and Related Observable Over-generation Errors. In this work, we present our solution to this problem, which capitalizes on the variability of stochastically… ▽ More Identification of hallucination spans in black-box language model generated text is essential for applications in the real world. A recent attempt at this direction is SemEval-2025 Task 3, Mu-SHROOM-a Multilingual Shared Task on Hallucinations and Related Observable Over-generation Errors. In this work, we present our solution to this problem, which capitalizes on the variability of stochastically-sampled responses in order to identify hallucinated spans. Our hypothesis is that if a language model is certain of a fact, its sampled responses will be uniform, while hallucinated facts will yield different and conflicting results. We measure this divergence through entropy-based analysis, allowing for accurate identification of hallucinated segments. Our method is not dependent on additional training and hence is cost-effective and adaptable. In addition, we conduct extensive hyperparameter tuning and perform error analysis, giving us crucial insights into model behavior. △ Less

Submitted 23 May, 2025; originally announced May 2025.

Graph Transformer-Based Flood Susceptibility Mapping: Application to the French Riviera and Railway Infrastructure Under Climate Change

Authors: Sreenath Vemula, Filippo Gatti, Pierre Jehel

Abstract: Increasing flood frequency and severity due to climate change threatens infrastructure and demands improved susceptibility mapping techniques. While traditional machine learning (ML) approaches are widely used, they struggle to capture spatial dependencies and poor boundary delineation between susceptibility classes. This study introduces the first application of a graph transformer (GT) architect… ▽ More Increasing flood frequency and severity due to climate change threatens infrastructure and demands improved susceptibility mapping techniques. While traditional machine learning (ML) approaches are widely used, they struggle to capture spatial dependencies and poor boundary delineation between susceptibility classes. This study introduces the first application of a graph transformer (GT) architecture for flood susceptibility mapping to the flood-prone French Riviera (e.g., 2020 Storm Alex) using topography, hydrology, geography, and environmental data. GT incorporates watershed topology using Laplacian positional encoders (PEs) and attention mechanisms. The developed GT model has an AUC-ROC (0.9739), slightly lower than XGBoost (0.9853). However, the GT model demonstrated better clustering and delineation with a higher Moran's I value (0.6119) compared to the random forest (0.5775) and XGBoost (0.5311) with p-value lower than 0.0001. Feature importance revealed a striking consistency across models, with elevation, slope, distance to channel, and convergence index being the critical factors. Dimensionality reduction on Laplacian PEs revealed partial clusters, indicating they could capture spatial information; however, their importance was lower than flood factors. Since climate and land use changes aggravate flood risk, susceptibility maps are developed for the 2050 year under different Representative Concentration Pathways (RCPs) and railway track vulnerability is assessed. All RCP scenarios revealed increased area across susceptibility classes, except for the very low category. RCP 8.5 projections indicate that 17.46% of the watershed area and 54% of railway length fall within very-high susceptible zones, compared to 6.19% and 35.61%, respectively, under current conditions. The developed maps can be integrated into a multi-hazard framework. △ Less

Submitted 31 March, 2025; originally announced April 2025.

Comments: Submitted to Science of Total Environment journal

DisBeaNet: A Deep Neural Network to augment Unmanned Surface Vessels for maritime situational awareness

Authors: Srikanth Vemula, Eulises Franco, Michael Frye

Abstract: Intelligent detection and tracking of the vessels on the sea play a significant role in conducting traffic avoidance in unmanned surface vessels(USV). Current traffic avoidance software relies mainly on Automated Identification System (AIS) and radar to track other vessels to avoid collisions and acts as a typical perception system to detect targets. However, in a contested environment, emitting r… ▽ More Intelligent detection and tracking of the vessels on the sea play a significant role in conducting traffic avoidance in unmanned surface vessels(USV). Current traffic avoidance software relies mainly on Automated Identification System (AIS) and radar to track other vessels to avoid collisions and acts as a typical perception system to detect targets. However, in a contested environment, emitting radar energy also presents the vulnerability to detection by adversaries. Deactivating these Radiofrequency transmitting sources will increase the threat of detection and degrade the USV's ability to monitor shipping traffic in the vicinity. Therefore, an intelligent visual perception system based on an onboard camera with passive sensing capabilities that aims to assist USV in addressing this problem is presented in this paper. This paper will present a novel low-cost vision perception system for detecting and tracking vessels in the maritime environment. This novel low-cost vision perception system is introduced using the deep learning framework. A neural network, DisBeaNet, can detect vessels, track, and estimate the vessel's distance and bearing from the monocular camera. The outputs obtained from this neural network are used to determine the latitude and longitude of the identified vessel. △ Less

Submitted 17 May, 2024; v1 submitted 9 May, 2024; originally announced May 2024.

Learning Analytics Dashboards for Advisors -- A Systematic Literature Review

Authors: Suchith Reddy Vemula, Marcia Moraes

Abstract: Learning Analytics Dashboard for Advisors is designed to provide data-driven insights and visualizations to support advisors in their decision-making regarding student academic progress, engagement, targeted support, and overall success. This study explores the current state of the art in learning analytics dashboards, focusing on specific requirements for advisors. By examining existing literatur… ▽ More Learning Analytics Dashboard for Advisors is designed to provide data-driven insights and visualizations to support advisors in their decision-making regarding student academic progress, engagement, targeted support, and overall success. This study explores the current state of the art in learning analytics dashboards, focusing on specific requirements for advisors. By examining existing literature and case studies, this research investigates the key features and functionalities essential for an effective learning analytics dashboard tailored to advisor needs. This study also aims to provide a comprehensive understanding of the landscape of learning analytics dashboards for advisors, offering insights into the advancements, opportunities, and challenges in their development by synthesizing the current trends from a total of 21 research papers used for analysis. The findings will contribute to the design and implementation of new features in learning analytics dashboards that empower advisors to provide proactive and individualized support, ultimately fostering student retention and academic success. △ Less

Submitted 17 January, 2024; originally announced February 2024.