arXiv subjects

Saptarsi Goswami

Publications and source records attributed to Saptarsi Goswami.

10 recordsLinked to original sources

arXiv2026

MMTClinic: Multimodal, Multilingual Time Series Question Answering and Reasoning Benchmark for Clinical Domain

Time-series data in clinical settings is crucial for capturing dynamic changes in a patient's health over time, enabling timely diagnosis, personalized treatment, and early detection of critical events. However, the development of clinically reliable and linguistically inclusive medical AI systems remains a significant challenge, primarily due to the lack of multimodal, multilingual, and time-series-grounded benchmarks that reflect the complexity of real-world clinical scenarios. To fill this gap, we present MMTClinic, a benchmark designed to evaluate large language models (LLMs) on complex reasoning and question-answering tasks involving clinical time-series. MMTClinic combines text, medical images, and multivariate physiological signals and includes 30,000 QA pairs (15,000 multiple choice questions (MCQs) and 15,000 open-ended questions) across five languages: English, Hindi, Bengali, Marathi, and Tamil. These questions cover three important clinical tasks---mortality prediction, heart rate forecasting, and SOFA score estimation. We evaluate 13 state-of-the-art LLMs in zero-shot, few-shot, and chain-of-thought settings. Our evaluation reveals notable differences in model performance across tasks, languages, and modalities, highlighting current limitations in clinical reasoning capabilities. MMTClinic provides a valuable resource for advancing multilingual, multimodal, and time-series-aware medical AI research. The dataset will be made publicly available on successful acceptance of the work.

cs.CL↗

arXiv2025

A Data Science Approach to Calcutta High Court Judgments: An Efficient LLM and RAG-powered Framework for Summarization and Similar Cases Retrieval

The judiciary, as one of democracy's three pillars, is dealing with a rising amount of legal issues, needing careful use of judicial resources. This research presents a complex framework that leverages Data Science methodologies, notably Large Language Models (LLM) and Retrieval-Augmented Generation (RAG) techniques, to improve the efficiency of analyzing Calcutta High Court verdicts. Our framework focuses on two key aspects: first, the creation of a robust summarization mechanism that distills complex legal texts into concise and coherent summaries; and second, the development of an intelligent system for retrieving similar cases, which will assist legal professionals in research and decision making. By fine-tuning the Pegasus model using case head note summaries, we achieve significant improvements in the summarization of legal cases. Our two-step summarizing technique preserves crucial legal contexts, allowing for the production of a comprehensive vector database for RAG. The RAG-powered framework efficiently retrieves similar cases in response to user queries, offering thorough overviews and summaries. This technique not only improves legal research efficiency, but it also helps legal professionals and students easily acquire and grasp key legal information, benefiting the overall legal scenario.

cs.IR↗

arXiv2024

BEExformer: A Fast Inferencing Binarized Transformer with Early Exits

Large Language Models (LLMs) based on transformers achieve cutting-edge results on a variety of applications. However, their enormous size and processing requirements hinder deployment on constrained resources. To enhance efficiency, binarization and Early Exit (EE) have proved to be effective solutions. However, binarization may lead to performance loss as reduced precision affects gradient estimation and parameter updates. Besides, research on EE mechanisms is still in its early stages. To address these challenges, we introduce Binarized Early Exit Transformer (BEExformer), a first-of-its-kind selective learning-based transformer integrating Binarization-Aware Training (BAT) with EE for efficient and fast textual inference. Each transformer block has an integrated Selective-Learn Forget Network (SLFN) to enhance contextual retention while eliminating irrelevant information. The BAT employs a differentiable second-order approximation to the sign function, enabling gradient computation that captures both the sign and magnitude of the weights. This aids in 21.30 times reduction in model size. The EE mechanism hinges on fractional reduction in entropy among intermediate transformer blocks with soft-routing loss estimation. This accelerates inference by reducing FLOPs by 52.27% and even improves accuracy by 3.22% by resolving the "overthinking" problem inherent in deep networks. Extensive evaluation through comparison with the SOTA methods and various ablations across nine datasets covering multiple NLP tasks demonstrates its Pareto-optimal performance-efficiency trade-off.

cs.CL↗

arXiv2024

A Novel Denoising Technique and Deep Learning Based Hybrid Wind Speed Forecasting Model for Variable Terrain Conditions

Wind flow can be highly unpredictable and can suffer substantial fluctuations in speed and direction due to the shape and height of hills, mountains, and valleys, making accurate wind speed (WS) forecasting essential in complex terrain. This paper presents a novel and adaptive model for short-term forecasting of WS. The paper's key contributions are as follows: (a) The Partial Auto Correlation Function (PACF) is utilised to minimise the dimension of the set of Intrinsic Mode Functions (IMF), hence reducing training time; (b) The sample entropy (SampEn) was used to calculate the complexity of the reduced set of IMFs. The proposed technique is adaptive since a specific Deep Learning (DL) model-feature combination was chosen based on complexity; (c) A novel bidirectional feature-LSTM framework for complicated IMFs has been suggested, resulting in improved forecasting accuracy; (d) The proposed model shows superior forecasting performance compared to the persistence, hybrid, Ensemble empirical mode decomposition (EEMD), and Variational Mode Decomposition (VMD)-based deep learning models. It has achieved the lowest variance in terms of forecasting accuracy between simple and complex terrain conditions 0.70%. Dimension reduction of IMF's and complexity-based model-feature selection helps reduce the training time by 68.77% and improve forecasting quality by 58.58% on average.

cs.LG↗

arXiv2024

TexIm FAST: Text-to-Image Representation for Semantic Similarity Evaluation using Transformers

One of the principal objectives of Natural Language Processing (NLP) is to generate meaningful representations from text. Improving the informativeness of the representations has led to a tremendous rise in the dimensionality and the memory footprint. It leads to a cascading effect amplifying the complexity of the downstream model by increasing its parameters. The available techniques cannot be applied to cross-modal applications such as text-to-image. To ameliorate these issues, a novel Text-to-Image methodology for generating fixed-length representations through a self-supervised Variational Auto-Encoder (VAE) for semantic evaluation applying transformers (TexIm FAST) has been proposed in this paper. The pictorial representations allow oblivious inference while retaining the linguistic intricacies, and are potent in cross-modal applications. TexIm FAST deals with variable-length sequences and generates fixed-length representations with over 75% reduced memory footprint. It enhances the efficiency of the models for downstream tasks by reducing its parameters. The efficacy of TexIm FAST has been extensively analyzed for the task of Semantic Textual Similarity (STS) upon the MSRPC, CNN/ Daily Mail, and XSum data-sets. The results demonstrate 6% improvement in accuracy compared to the baseline and showcase its exceptional ability to compare disparate length sequences such as a text with its summary.

cs.CL↗

arXiv2024

A Survey of Transformer-based Language Models with Focus on Efficiency

The emergence of Transformer-based Large Language Models (LLMs) has substantially augmented the capabilities of Natural Language Processing (NLP), thereby intensifying the demand for computational resources. Therefore, enhancing efficiency based on factors like computational requirements, energy consumption, carbon footprint and financial cost has become a vital area of research. This motivates us to conduct a survey on Transformer-based LLMs in NLP from the perspective of efficiency. In this survey of 312 articles, the efficiency-improvement endeavors have been systematically discussed targeting various aspects such as data curation, model design, model downsizing, and dynamic inferencing. This has been augmented with efficiency considerations in model adaptation strategies like pre-training, fine-tuning, prompt-engineering and Retrieval-Augmented Generation (RAG). Furthermore, a statistical analysis followed by an in-depth evaluation of the efficiency and efficacy of more than 30 renowned NLP models has been performed on 13 evaluation benchmarks. This paper offers valuable insights for researchers, professionals as well as scholars, and explores the trend of research toward sustainable practices in NLP.

cs.CL↗

arXiv2016

A review on application of data mining techniques to combat natural disasters

Thousands of human lives are lost every year around the globe, apart from significant damage on property, animal life, etc., due to natural disasters (e.g., earthquake, flood, tsunami, hurricane and other storms, landslides, cloudburst, heat wave, forest fire). In this paper, we focus on reviewing the application of data mining and analytical techniques designed so far for (i) prediction, (ii) detection, and (iii) development of appropriate disaster management strategy based on the collected data from disasters. A detailed description of availability of data from geological observatories (seismological, hydrological), satellites, remote sensing and newer sources like social networking sites as twitter is presented. An extensive and in-depth literature study on current techniques for disaster prediction, detection and management has been done and the results are summarized according to various types of disasters. Finally a framework for building a disaster management database for India hosted on open source Big Data platform like Hadoop in a phased manner has been proposed. The study has special focus on India which ranks among top five counties in terms of absolute number of the loss of human life.

cs.CY↗

arXiv2012

Quartile Clustering: A quartile based technique for Generating Meaningful Clusters

Clustering is one of the main tasks in exploratory data analysis and descriptive statistics where the main objective is partitioning observations in groups. Clustering has a broad range of application in varied domains like climate, business, information retrieval, biology, psychology, to name a few. A variety of methods and algorithms have been developed for clustering tasks in the last few decades. We observe that most of these algorithms define a cluster in terms of value of the attributes, density, distance etc. However these definitions fail to attach a clear meaning/semantics to the generated clusters. We argue that clusters having understandable and distinct semantics defined in terms of quartiles/halves are more appealing to business analysts than the clusters defined by data boundaries or prototypes. On the samepremise, we propose our new algorithm named as quartile clustering technique. Through a series of experiments we establish efficacy of this algorithm. We demonstrate that the quartile clustering technique adds clear meaning to each of the clusters compared to K-means. We use DB Index to measure goodness of the clusters and show our method is comparable to EM (Expectation Maximization), PAM (Partition around Medoid) and K Means. We have explored its capability in detecting outlier and the benefit of added semantics. We discuss some of the limitations in its present form and also provide a rough direction in addressing the issue of merging the generated clusters.

cs.DB↗

arXiv2012

Outlier Detection Techniques for SQL and ETL Tuning

RDBMS is the heart for both OLTP and OLAP types of applications. For both types of applications thousands of queries expressed in terms of SQL are executed on daily basis. All the commercial DBMS engines capture various attributes in system tables about these executed queries. These queries need to conform to best practices and need to be tuned to ensure optimal performance. While we use checklists, often tools to enforce the same, a black box technique on the queries for profiling, outlier detection is not employed for a summary level understanding. This is the motivation of the paper, as this not only points out to inefficiencies built in the system, but also has the potential to point evolving best practices and inappropriate usage. Certainly this can reduce latency in information flow and optimal utilization of hardware and software capacity. In this paper we start with formulating the problem. We explore four outlier detection techniques. We apply these techniques over rich corpora of production queries and analyze the results. We also explore benefit of an ensemble approach. We conclude with future courses of action. The same philosophy we have used for optimization of extraction, transform, load (ETL) jobs in one of our previous work. We give a brief introduction of the same in section four.

cs.DB↗

arXiv2012

Outlier detection from ETL Execution trace

Extract, Transform, Load (ETL) is an integral part of Data Warehousing (DW) implementation. The commercial tools that are used for this purpose captures lot of execution trace in form of various log files with plethora of information. However there has been hardly any initiative where any proactive analyses have been done on the ETL logs to improve their efficiency. In this paper we utilize outlier detection technique to find the processes varying most from the group in terms of execution trace. As our experiment was carried on actual production processes, any outlier we would consider as a signal rather than a noise. To identify the input parameters for the outlier detection algorithm we employ a survey among developer community with varied mix of experience and expertise. We use simple text parsing to extract these features from the logs, as shortlisted from the survey. Subsequently we applied outlier detection technique (Clustering based) on the logs. By this process we reduced our domain of detailed analysis from 500 logs to 44 logs (8 Percentage). Among the 5 outlier cluster, 2 of them are genuine concern, while the other 3 figure out because of the huge number of rows involved.

cs.DB↗