Publications

The Open Data Protocol (OData) provides a standardized approach for building and consuming RESTful APIs with rich query capabilities. Despite its power and maturity, OData adoption remains confined primarily to enterprise environments, particularly within Microsoft and SAP ecosystems. This paper analyzes the key barriers preventing wider OData adoption and introduces ODataX, an evolved version of the protocol designed to address these limitations. ODataX maintains backward compatibility with OData v4 while introducing progressive complexity disclosure through simplified query syntax, built-in performance guardrails via query cost estimation, and enhanced caching mechanisms.

We present a reproducibility study of the state-of-the-art neural architecture for sequence labeling proposed by Ma and Hovy (2016). The original BiLSTM-CNN-CRF model combines character-level representations via Convolutional Neural Networks (CNNs), word-level context modeling through Bi-directional Long Short-Term Memory networks (BiLSTMs), and structured prediction using Conditional Random Fields (CRFs). This end-to-end approach eliminates the need for hand-crafted features while achieving excellent performance on named entity recognition (NER) and part-of-speech (POS) tagging tasks. Our implementation successfully reproduces the key results, achieving 91.18% F1-score on CoNLL-2003 NER.

This paper applies Amdahl's Law to analyze the scalability characteristics of Large Language Model (LLM) completion endpoints. We examine the inherent parallelizability of LLM inference workloads, identifying the sequential and parallelizable components of the request processing pipeline. The analysis provides practical insights into the theoretical speedup limits of LLM serving systems and offers guidance for optimizing endpoint architectures to maximize throughput under concurrent load.

This paper presents an adaptive learning neural network approach for the segmentation and classification of heart sounds (phonocardiogram signals). The proposed method addresses the challenge of identifying normal and abnormal cardiac conditions through automated analysis of heart sound recordings. We develop a system that segments heart sounds into their fundamental components and classifies them using neural networks with adaptive learning rates, achieving robust performance across varying recording conditions.

Handwritten character recognition has long been an important area of research in pattern recognition. The complexity of the task varies among different languages due to similarity between characters, distinct shapes, and number of characters. In this paper, we explored the performance of Deep Belief Networks in the classification of Handwritten Tamil vowels, and conclusively compared the results obtained. The proposed method has shown satisfactory recognition accuracy in light of difficulties faced with regional languages such as similarity between characters and minute nuances that differentiate them.

Handwritten character recognition has long been an important area of research in pattern recognition. The complexity varies among different languages due to similarity between characters, distinct shapes, and number of characters. In this paper, we explored the performance of Convolutional Neural Networks and Deep Belief Networks in the classification of Handwritten Kannada numerals, and conclusively compared the results obtained. The proposed method has shown satisfactory recognition accuracy in light of difficulties faced with regional languages such as similarity between characters and minute nuances that differentiate them.