arXiv subjects

Uthayasanker Thayasivam

Publications and source records attributed to Uthayasanker Thayasivam.

At least 19 recordsLinked to original sources

arXiv2026

Embedding Graph Kernel Entropy (EGKE): Fine-grained Hallucination Detection for LLMs from Embeddings

Hallucination detection is critical for ensuring the reliability of Large Language Models (LLMs) in real-world deployments. While recent works, such as Semantic Entropy (SE) and Kernel Language Entropy (KLE), identify hallucinations by measuring semantic disagreement across multiple generated responses using Natural Language Inference (NLI) models, these methods depend on separately loaded NLI classifiers to estimate pairwise semantic similarity, introducing significant computational overhead. We propose Embedding Graph Kernel Entropy (EGKE), a variant of Kernel Language Entropy (KLE) that replaces its NLI-derived semantic graph with a Radial Basis Function (RBF) kernel applied to dense sentence embeddings generated by a BGE embedding model, while retaining KLE's graph Laplacian, heat kernel, and von Neumann entropy formulation unchanged. By computing similarities directly in a continuous embedding space, the proposed method provides a symmetric and computationally efficient alternative to NLI-based semantic comparison. We evaluate our approach across 30 experimental setups (covering multiple question-answer datasets and open-weight LLM families) using both AUROC and AUARC, and demonstrate that this RBF-based method achieves competitive hallucination detection performance against state-of-the-art methods.

cs.LG↗

arXiv2026

Do LLMs Understand Context? A Knowledge Graph-Based Evaluation Framework

While large language models (LLMs) have achieved remarkable linguistic capabilities, a profound question lingers at their core: do these models truly comprehend context or simply excel at pattern matching on an unprecedented scale? Contextual understanding in LLMs refers to the ability to correctly extract relevant information from a given context, integrate it into a coherent internal representation, and reason over it to produce factually consistent and contextually grounded responses. However, traditional methods such as BiLingual Evaluation Understudy (BLEU) and perplexity simply measure surface-level performance. This reveals a critical gap in question answering (QA), where responses must be contextually grounded rather than simply being memorized associations. To fill this void, we propose a novel knowledge graph (KG) based evaluation framework for LLM contextual understanding in QA. Central to this is Semantic Structural Similarity for KGs (S3KG), a hybrid similarity measure combining structural and semantic signals into a single score. In addition, a diagnostic analysis framework is developed to identify and categorize reasoning errors at the triplet level, enabling fine-grained analysis of model failures. Together, across nine benchmarks, S3KG achieves F1 gains of up to $+7.6$ points over the strongest baseline and AUROC up to $0.973$.

cs.AI↗

arXiv2026

Evaluation of Contextual Understanding in Large Language Models

Large Language Models (LLMs) demonstrate impressive performance across diverse NLP tasks, yet their ability to exhibit genuine contextual understanding remains uncertain. Traditional evaluation metrics such as perplexity, BiLingual Evaluation Understudy (BLEU), or surface-level accuracy fail to reveal how well LLMs extract, integrate, and reason over contextual information--a gap particularly critical in question answering, where models must align responses with contextually grounded knowledge rather than memorized associations. We propose a novel knowledge graph-based evaluation framework introducing Semantic Structural Similarity for KGs (S3KG), a hybrid similarity measure integrating structural and semantic similarity into a continuous evaluation score, alongside a diagnostic framework for categorizing reasoning errors. To validate this pipeline, we evaluate S3KG against established metrics on a curated question-answer (QA) benchmark, demonstrating its effectiveness in measuring correctness, faithfulness, and interpretability in LLM-generated responses.

cs.CL↗

arXiv2026

EnSiTa - A Trilingual Multi-Domain Parallel Dataset and Benchmark for Domain-Specific Machine Translation

Machine Translation (MT) for low-resource languages remains far behind that of high-resource languages, and the gap is widest in specialised domains, where parallel data is scarce or entirely absent. We present EnSiTa, a trilingual multi-domain parallel dataset and benchmark for English, Sinhala and Tamil. EnSiTa provides human post-edited training data for seven domains, plus manually translated test sets for those and one additional domain, all produced by professional translators under a multi-year, rigorously quality-controlled process. Using this dataset, we conduct an extensive study of domain-specific MT for all six language directions, fine-tuning a from-scratch Transformer, a pre-trained translation model (NLLB-600M), and decoder-only LLMs (Gemma 3 family, 1B-12B, and TranslateGemma) across training-data sizes, model scales, and in-domain, cross-domain, multilingual and multi-domain settings. To the best of our knowledge, this is the most extensive systematically documented multi-domain parallel data creation and benchmarking effort for low-resource MT. Our data and models will be publicly released.

cs.CL↗

arXiv2026

OasisSimp: An Open-source Asian-English Sentence Simplification Dataset

Sentence simplification aims to make complex text more accessible by reducing linguistic complexity while preserving the original meaning. However, progress in this area remains limited for mid-resource and low-resource languages due to the scarcity of high-quality data. To address this gap, we introduce the OasisSimp dataset, a multilingual dataset for sentence-level simplification covering five languages: English, Sinhala, Tamil, Pashto, and Thai. Among these, no prior sentence simplification datasets exist for Thai, Pashto, and Tamil, while limited data is available for Sinhala. Each language simplification dataset was created by trained annotators who followed detailed guidelines to simplify sentences while maintaining meaning, fluency, and grammatical correctness. We evaluate eight open-weight multilingual Large Language Models (LLMs) on the OasisSimp dataset and observe substantial performance disparities between high-resource and low-resource languages, highlighting the simplification challenges in multilingual settings. The OasisSimp dataset thus provides both a valuable multilingual resource and a challenging benchmark, revealing the limitations of current LLM-based simplification methods and paving the way for future research in low-resource sentence simplification. The dataset is available at https://OasisSimpDataset.github.io/.

cs.CL↗

arXiv2025

Adapting Multimodal Foundation Models for Few-Shot Learning: A Comprehensive Study on Contrastive Captioners

Large-scale multimodal foundation models, particularly Contrastive Captioners (CoCa), have achieved state-of-the-art results by unifying contrastive alignment with generative captioning. While zero-shot transfer capabilities are well-documented, the adaptation of these generative-contrastive hybrids to downstream tasks with extreme data scarcity (few-shot learning) remains under-explored. Existing literature predominantly focuses on dual-encoder architectures like CLIP, leaving a gap in understanding how CoCa's distinct latent space responds to parameter-efficient fine-tuning (PEFT). This paper presents a comprehensive empirical study on adapting the CoCa visual backbone for few-shot image classification. We systematically evaluate a hierarchy of strategies, ranging from training-free hybrid prototyping to deep parameter adaptation via Low-Rank Adaptation (LoRA). First, we identify an "augmentation divergence": while strong data augmentation degrades the performance of linear probing in low-shot settings, it is essential for stabilizing LoRA fine-tuning. We also demonstrate that hybrid objectives incorporating Supervised Contrastive (SupCon) loss yield consistent performance improvements over standard Cross-Entropy across varying shot counts. Crucially, we characterize the sensitivity of training configurations to data scarcity, providing empirical reference settings for scaling regularization, rank, and sampling strategies to facilitate the efficient adaptation of generative-contrastive foundation models.

cs.CV↗

arXiv2025

Team-Aware Football Player Tracking with SAM: An Appearance-Based Approach to Occlusion Recovery

Football player tracking is challenged by frequent occlusions, similar appearances, and rapid motion in crowded scenes. This paper presents a lightweight SAM-based tracking method combining the Segment Anything Model (SAM) with CSRT trackers and jersey color-based appearance models. We propose a team-aware tracking system that uses SAM for precise initialization and HSV histogram-based re-identification to improve occlusion recovery. Our evaluation measures three dimensions: processing speed (FPS and memory), tracking accuracy (success rate and box stability), and robustness (occlusion recovery and identity consistency). Experiments on football video sequences show that the approach achieves 7.6-7.7 FPS with stable memory usage (~1880 MB), maintaining 100 percent tracking success in light occlusions and 90 percent in crowded penalty-box scenarios with 5 or more players. Appearance-based re-identification recovers 50 percent of heavy occlusions, demonstrating the value of domain-specific cues. Analysis reveals key trade-offs: the SAM + CSRT combination provides consistent performance across crowd densities but struggles with long-term occlusions where players leave the frame, achieving only 8.66 percent re-acquisition success. These results offer practical guidelines for deploying football tracking systems under resource constraints, showing that classical tracker-based methods work well with continuous visibility but require stronger re-identification mechanisms for extended absences.

cs.CV↗

arXiv2025

TBT-Former: Learning Temporal Boundary Distributions for Action Localization

Temporal Action Localization (TAL) remains a fundamental challenge in video understanding, aiming to identify the start time, end time, and category of all action instances within untrimmed videos. While recent single-stage, anchor-free models like ActionFormer have set a high standard by leveraging Transformers for temporal reasoning, they often struggle with two persistent issues: the precise localization of actions with ambiguous or "fuzzy" temporal boundaries and the effective fusion of multi-scale contextual information. In this paper, we introduce the Temporal Boundary Transformer (TBT-Former), a new architecture that directly addresses these limitations. TBT-Former enhances the strong ActionFormer baseline with three core contributions: (1) a higher-capacity scaled Transformer backbone with an increased number of attention heads and an expanded Multi-Layer Perceptron (MLP) dimension for more powerful temporal feature extraction; (2) a cross-scale feature pyramid network (FPN) that integrates a top-down pathway with lateral connections, enabling richer fusion of high-level semantics and low-level temporal details; and (3) a novel boundary distribution regression head. Inspired by the principles of Generalized Focal Loss (GFL), this new head recasts the challenging task of boundary regression as a more flexible probability distribution learning problem, allowing the model to explicitly represent and reason about boundary uncertainty. Within the paradigm of Transformer-based architectures, TBT-Former advances the formidable benchmark set by its predecessors, establishing a new level of performance on the highly competitive THUMOS14 and EPIC-Kitchens 100 datasets, while remaining competitive on the large-scale ActivityNet-1.3. Our code is available at https://github.com/aaivu/In21-S7-CS4681-AML-Research-Projects/tree/main/projects/210536K-Multi-Modal-Learning_Video-Understanding

cs.CV↗

arXiv2025

Pushing the Boundaries of Interpretability: Incremental Enhancements to the Explainable Boosting Machine

The widespread adoption of complex machine learning models in high-stakes domains has brought the "black-box" problem to the forefront of responsible AI research. This paper aims at addressing this issue by improving the Explainable Boosting Machine (EBM), a state-of-the-art glassbox model that delivers both high accuracy and complete transparency. The paper outlines three distinct enhancement methodologies: targeted hyperparameter optimization with Bayesian methods, the implementation of a custom multi-objective function for fairness for hyperparameter optimization, and a novel self-supervised pre-training pipeline for cold-start scenarios. All three methodologies are evaluated across standard benchmark datasets, including the Adult Income, Credit Card Fraud Detection, and UCI Heart Disease datasets. The analysis indicates that while the tuning process yielded marginal improvements in the primary ROC AUC metric, it led to a subtle but important shift in the model's decision-making behavior, demonstrating the value of a multi-faceted evaluation beyond a single performance score. This work is positioned as a critical step toward developing machine learning systems that are not only accurate but also robust, equitable, and transparent, meeting the growing demands of regulatory and ethical compliance.

cs.LG↗

arXiv2025

TS2Vec-Ensemble: An Enhanced Self-Supervised Framework for Time Series Forecasting

Self-supervised representation learning, particularly through contrastive methods like TS2Vec, has advanced the analysis of time series data. However, these models often falter in forecasting tasks because their objective functions prioritize instance discrimination over capturing the deterministic patterns, such as seasonality and trend, that are critical for accurate prediction. This paper introduces TS2Vec-Ensemble, a novel hybrid framework designed to bridge this gap. Our approach enhances the powerful, implicitly learned dynamics from a pretrained TS2Vec encoder by fusing them with explicit, engineered time features that encode periodic cycles. This fusion is achieved through a dual-model ensemble architecture, where two distinct regression heads -- one focused on learned dynamics and the other on seasonal patterns -- are combined using an adaptive weighting scheme. The ensemble weights are optimized independently for each forecast horizon, allowing the model to dynamically prioritize short-term dynamics or long-term seasonality as needed. We conduct extensive experiments on the ETT benchmark datasets for both univariate and multivariate forecasting. The results demonstrate that TS2Vec-Ensemble consistently and significantly outperforms the standard TS2Vec baseline and other state-of-the-art models, validating our hypothesis that a hybrid of learned representations and explicit temporal priors is a superior strategy for long-horizon time series forecasting.

cs.LG↗

arXiv2025

Enhancing Multi-Label Thoracic Disease Diagnosis with Deep Ensemble-Based Uncertainty Quantification

The utility of deep learning models, such as CheXNet, in high stakes clinical settings is fundamentally constrained by their purely deterministic nature, failing to provide reliable measures of predictive confidence. This project addresses this critical gap by integrating robust Uncertainty Quantification (UQ) into a high performance diagnostic platform for 14 common thoracic diseases on the NIH ChestX-ray14 dataset. Initial architectural development failed to stabilize performance and calibration using Monte Carlo Dropout (MCD), yielding an unacceptable Expected Calibration Error (ECE) of 0.7588. This technical failure necessitated a rigorous architectural pivot to a high diversity, 9-member Deep Ensemble (DE). This resulting DE successfully stabilized performance and delivered superior reliability, achieving a State-of-the-Art (SOTA) average Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.8559 and an average F1 Score of 0.3857. Crucially, the DE demonstrated superior calibration (Mean ECE of 0.0728 and Negative Log-Likelihood (NLL) of 0.1916) and enabled the reliable decomposition of total uncertainty into its Aleatoric (irreducible data noise) and Epistemic (reducible model knowledge) components, with a mean Epistemic Uncertainty (EU) of 0.0240. These results establish the Deep Ensemble as a trustworthy and explainable platform, transforming the model from a probabilistic tool into a reliable clinical decision support system.

cs.CV↗

arXiv2025

Attack-Specialized Deep Learning with Ensemble Fusion for Network Anomaly Detection

The growing scale and sophistication of cyberattacks pose critical challenges to network security, particularly in detecting diverse intrusion types within imbalanced datasets. Traditional intrusion detection systems (IDS) often struggle to maintain high accuracy across both frequent and rare attacks, leading to increased false negatives for minority classes. To address this, we propose a hybrid anomaly detection framework that integrates specialized deep learning models with an ensemble meta-classifier. Each model is trained to detect a specific attack category, enabling tailored learning of class-specific patterns, while their collective outputs are fused by a Random Forest meta-classifier to improve overall decision reliability. The framework is evaluated on the NSL-KDD benchmark, demonstrating superior performance in handling class imbalance compared to conventional monolithic models. Results show significant improvements in precision, recall, and F1-score across all attack categories, including rare classes such as User to Root (U2R). The proposed system achieves near-perfect detection rates with minimal false alarms, highlighting its robustness and generalizability. This work advances the design of intrusion detection systems by combining specialization with ensemble learning, providing an effective and scalable solution for safeguarding modern networks.

cs.CR↗

arXiv2025

A Novel Multi-branch ConvNeXt Architecture for Identifying Subtle Pathological Features in CT Scans

Intelligent analysis of medical imaging plays a crucial role in assisting clinical diagnosis, especially for identifying subtle pathological features. This paper introduces a novel multi-branch ConvNeXt architecture designed specifically for the nuanced challenges of medical image analysis. While applied here to the specific problem of COVID-19 diagnosis, the methodology offers a generalizable framework for classifying a wide range of pathologies from CT scans. The proposed model incorporates a rigorous end-to-end pipeline, from meticulous data preprocessing and augmentation to a disciplined two-phase training strategy that leverages transfer learning effectively. The architecture uniquely integrates features extracted from three parallel branches: Global Average Pooling, Global Max Pooling, and a new Attention-weighted Pooling mechanism. The model was trained and validated on a combined dataset of 2,609 CT slices derived from two distinct datasets. Experimental results demonstrate a superior performance on the validation set, achieving a final ROC-AUC of 0.9937, a validation accuracy of 0.9757, and an F1-score of 0.9825 for COVID-19 cases, outperforming all previously reported models on this dataset. These findings indicate that a modern, multi-branch architecture, coupled with careful data handling, can achieve performance comparable to or exceeding contemporary state-of-the-art models, thereby proving the efficacy of advanced deep learning techniques for robust medical diagnostics.

cs.CV↗

arXiv2025

A Framework to Prevent Biometric Data Leakage in the Immersive Technologies Domain

Doubtlessly, the immersive technologies have potential to ease people's life and uplift economy, however the obvious data privacy risks cannot be ignored. For example, a participant wears a 3D headset device which detects participant's head motion to track the pose of participant's head to match the orientation of camera with participant's eyes positions in the real-world. In a preliminary study, researchers have proved that the voice command features on such headsets could lead to major privacy leakages. By analyzing the facial dynamics captured with the motion sensors, the headsets suffer security vulnerabilities revealing a user's sensitive speech without user's consent. The psychography data (such as voice command features, facial dynamics, etc.) is sensitive data and it should not be leaked out of the device without users consent else it is a privacy breach. To the best of our literature review, the work done in this particular research problem is very limited. Motivated from this, we develop a simple technical framework to mitigate sensitive data (or biometric data) privacy leaks in immersive technology domain. The performance evaluation is conducted in a robust way using six data sets, to show that the proposed solution is effective and feasible to prevent this issue.

cs.CR↗

arXiv2024

GPS-2-GTFS: A Python package to process and transform raw GPS data of public transit to GTFS format

The gps2gtfs package addresses a critical need for converting raw Global Positioning System (GPS) trajectory data from public transit vehicles into the widely used GTFS (General Transit Feed Specification) format. This transformation enables various software applications to efficiently utilize real-time transit data for purposes such as tracking, scheduling, and arrival time prediction. Developed in Python, gps2gtfs employs techniques like geo-buffer mapping, parallel processing, and data filtering to manage challenges associated with raw GPS data, including high volume, discontinuities, and localization errors. This open-source package, available on GitHub and PyPI, enhances the development of intelligent transportation solutions and fosters improved public transit systems globally.

cs.CY↗

arXiv2023

FlowChroma -- A Deep Recurrent Neural Network for Video Colorization

We develop an automated video colorization framework that minimizes the flickering of colors across frames. If we apply image colorization techniques to successive frames of a video, they treat each frame as a separate colorization task. Thus, they do not necessarily maintain the colors of a scene consistently across subsequent frames. The proposed solution includes a novel deep recurrent encoder-decoder architecture which is capable of maintaining temporal and contextual coherence between consecutive frames of a video. We use a high-level semantic feature extractor to automatically identify the context of a scenario including objects, with a custom fusion layer that combines the spatial and temporal features of a frame sequence. We demonstrate experimental results, qualitatively showing that recurrent neural networks can be successfully used to improve color consistency in video colorization.

cs.CV↗

arXiv2022

Adapting the Tesseract Open-Source OCR Engine for Tamil and Sinhala Legacy Fonts and Creating a Parallel Corpus for Tamil-Sinhala-English

Most low-resource languages do not have the necessary resources to create even a substantial monolingual corpus. These languages may often be found in government proceedings but mainly in Portable Document Format (PDF) that contains legacy fonts. Extracting text from these documents to create a monolingual corpus is challenging due to legacy font usage and printer-friendly encoding, which are not optimized for text extraction. Therefore, we propose a simple, automatic, and novel idea that can scale for Tamil, Sinhala, English languages, and many documents along with parallel corpora. Since Tamil and Sinhala are Low-Resource Languages, we improved the performance of Tesseract by employing LSTM-based training on more than 20 legacy fonts to recognize printed characters in these languages. Especially, our model detects code-mixed text, numbers, and special characters from the printed document. It is shown that this approach can reduce the character-level error rate of Tesseract from 6.03 to 2.61 for Tamil (-3.42% relative change) and 7.61 to 4.74 for Sinhala (-2.87% relative change), as well as the word-level error rate from 39.68 to 20.61 for Tamil (-19.07% relative change) and 35.04 to 26.58 for Sinhala (-8.46% relative change) on the test set. Also, our newly created parallel corpus consists of 185.4k, 168.9k, and 181.04k sentences and 2.11M, 2.22M, and 2.33M Words in Tamil, Sinhala, and English respectively. This study shows that fine-tuning Tesseract models on multiple new fonts help to understand the texts and enhances the performance of the OCR. We made newly trained models and the source code for fine-tuning Tesseract, freely available.

cs.CL↗

arXiv2022

Improving English to Sinhala Neural Machine Translation using Part-of-Speech Tag

The performance of Neural Machine Translation (NMT) depends significantly on the size of the available parallel corpus. Due to this fact, low resource language pairs demonstrate low translation performance compared to high resource language pairs. The translation quality further degrades when NMT is performed for morphologically rich languages. Even though the web contains a large amount of information, most people in Sri Lanka are unable to read and understand English properly. Therefore, there is a huge requirement of translating English content to local languages to share information among locals. Sinhala language is the primary language in Sri Lanka and building an NMT system that can produce quality English to Sinhala translations is difficult due to the syntactic divergence between these two languages under low resource constraints. Thus, in this research, we explore effective methods of incorporating Part of Speech (POS) tags to the Transformer input embedding and positional encoding to further enhance the performance of the baseline English to Sinhala neural machine translation model.

cs.CL↗