Dissertation and Thesis

Permanent URI for this communityhttps://dspace.isical.ac.in:4000/handle/10263/2146

Browse

Now showing 1 - 20 of 1346

2.5D Dual-Encoder U-Net for Lesion Segmentation in Chest CT Scans
(Indian Statistical Institute, Kolkata, 2025-06) Mukkara, Jagannath
Accurate segmentation of lesions in chest CT scans plays a vital role in diagnosing and monitoring pulmonary diseases such as COVID-19. In this, we introduce a novel 2.5D[1] dual-encoder U-Net model[2] that utilizes both the central slice and its neighboring slices to improve segmentation accuracy while keeping computational demands manageable. Our model incorporates residual connections[3] and feature fusion[4] to effectively merge multi-slice contextual information, overcoming the limitations found in traditional 2D and 3D methods. To ensure a reliable evaluation and avoid data leakage, we used patient-level data splitting. We validate our approach on a carefully curated chest CT dataset, showing enhanced segmentation performance and better generalization compared to standard U-Net models. Through extensive experiments, including ablation studies and visualizations, we demonstrate the advantages of combining 2.5D learning with a dual-encoder architecture for medical image segmentation tasks.
A Regression Tree Framework for Denoising and Monitoring of Image Data
(Indian Statistical Institute, 2026-07-02) Basak, Subhasish
The proliferation of advanced image acquisition technologies has led to the routine collection of large-scale image data across numerous scientific domains. This widespread reliance on image data accentuates the imperative to develop robust and efficient imaging techniques, which are essential for supporting modern applications across various scientific and industrial domains. This dissertation focuses on the development and analysis of methods for image denoising and image monitoring, two fundamental tasks in modern image analysis. A wide array of image denoising techniques exists in the literature, each tailored to handle specific types of noise or structural characteristics. However, no single method proves universally optimal, as each comes with its own advantages and trade-offs. In the first part of the dissertation, different configurations of local neighbourhoods are investigated, and an adaptive framework is proposed that combines these with local clustering-based smoothing to effectively harness the advantages of both methodologies. The dissertation then introduces a regression tree-based framework utilizing Oblique-axis Regression Trees (ORT) to estimate discontinuous regression functions in finite-dimensional spaces and applies this methodology to achieve effective image denoising. Due to an alternative set of assumptions on the underlying regression function, the overall structure of the proofs is substantially simpler than those typically found in the existing literature on regression trees. Finally, leveraging the ORT framework, the dissertation introduces an original approach to monitor drift patterns within an image sequence. Even though gradual temporal variations, known as drifts, are frequently observed in image sequences, drift monitoring remains an underexplored research area. This dissertation thus makes an effort to address that gap. Theoretical analysis and numerical studies, conducted on both simulated and real-world data, demonstrate the broad applicability and effectiveness of the proposed methods.
A Study of Prompt Tuning on Small Language Models(SLMs): A Controlled Benchmark and a Lightweight Instance-Aware Method
(2026-06-16) Sahith, Narkadamilli
Parameter-efficient fine-tuning (PEFT) adapts a frozen pre-trained language model by training only a small number of additional parameters. Among PEFT approaches, prompt tuning prepends trainable continuous vectors (soft prompts) to the input. A recurring finding in the literature is that prompt tuning is strongly scale dependent: it rivals full fine-tuning on very large models but lags on smaller ones. This dissertation studies prompt tuning specifically in the small-language-model (SLM) regime. We (i) re-implement a representative set of prompt-tuning methods—Prompt Tuning, P-Tuning v2, LoPT, DPT, DePT, ACCEPT, Residual Prompt Tuning, and PARA—within a single controlled harness, enabling a fair head-to-head comparison against full fine-tuning; (ii) propose IA-DePT, a lightweight instance-aware extension of Decomposed Prompt Tuning that conditions the short soft prompt on each input through a small, zero-initialised gate; and (iii) extend the benchmark beyond a single backbone and task, evaluating the full method suite on six backbone/task settings that span encoder–decoder (t5-small), encoder-only (BERT-base, RoBERTa-base, ELECTRA-small), and decoder-only (DistilGPT-2) architectures across the GLUE/SuperGLUE tasks RTE, WSC, CB, COPA, WiC, and MRPC. On RTE with t5-small, IA-DePT is the strongest parameter-efficient method in our benchmark (55.6% single-seed accuracy) and improves over its own base, DePT, by 6.5 points (53.6% vs. 47.1%, mean over three seeds) while adding only ≈16.9k parameters—a total trainable footprint of 0.05% of the backbone. Because the gate degrades exactly to DePT at initialisation, the comparison is a clean single-variable ablation. The cross-architecture study shows that the instance gate improves on DePT in five of the six settings on each setting’s primary metric (it ties or marginally regresses only on WiC, where every PEFT method sits at chance), so the benefit is broad but not universal. Our analysis characterises the accuracy/parameter trade-offs across method families, the strong effect of task difficulty on the small-model regime, and the role of instance-conditioning, including an honest discussion of why many prompt-tuning methods remain close to the chance baseline at this scale.
A Study of the SHA-2 Cryptographic Hash Family
(Indian Statistical Institute, Kolkata, 2009-02-01) Sanadhya Somitra Kumar
A Switch-Point-Aware Contrastive Approach to Sentiment Analysis of Hinglish Code-Mixed Text
(Indian Statistical Institute, 2026-06-15) Sahoo, Prasant Kumar
With the increasing use of social media in non-English-speaking regions, especially in India, people often use Romanized Hindi and English together in their online communication. In a single sentence, they frequently mix Romanized Hindi and English, creating code-mixed text. However, most multilingual transformer models are pre-trained primarily on monolingual data. As a result, NLP systems face challenges when processing code-mixed text, as a single word may be fragmented into meaningless subword pieces, making it difficult for the model to capture its semantic meaning accurately. In this dissertation, we propose a parameter efficient neural architecture consisting of three main components to address these challenges: First, there is a character-level CNN encoder, which handles spelling differences such as "nahi", "nahin", "nah", and "nai" through the chracter n-gram pattern. Next, there is a frozen XLM-R backbone(Conneau et al., 2019) , the top three layers, which are partly fine-tuned at a slower rate by which it provides rich cross lingual embeddings. Finally, there is a switch-point-aware bilingual gate that spots where the language label switches and blends two adapters using a learned gate weight.During training, it uses Supervised Contrastive Loss to learn better feature representations and Cross-Entropy Loss for classification. Since human annotators agreed on labels only 55% of the time, we use label smoothing to reflect this uncertainty and prevent the model from becoming overly confident in noisy labels. Evaluated on the SentiMix 2020 benchmark(Patwa et al., 2020), our proposed architecture achieves a weighted F1 score of 0.705, which outperforms the baseline model M-BERT (0.654 F1) and is comparable to fully fine-tuned transformer models while requiring only one-tenth of the trainable parameters.Adapter gate visualizations provide interpretable evidence that the gating mechanism captures linguistically meaningful codemixing structure. The architecture is designed to generalize to other code-mixed language pairs through its modular adapter design.
A1-homotopy types of A2 and A2 \ {(0, 0)}
(Indian Statistical Institute, Kolkata, 2024-12) Roy, Biman
Morel-Voevodsky developed A^1-homotopy theory which is a bridge between algebraic geometry and algebraic topology. In this thesis we study the A^1-connected component of a smooth variety in great detail. We have shown that the A^1-connected component of a smooth variety contains the information about the existence of affine lines in the variety. Using this and Miyanishi-Sugie's algebraic characterisation, we determine that the affine plane is the only A^1-contractible smooth affine surface over the field of characteristic zero. In the other part of the thesis, we studied the A^1-homotopy type of A^2-{(0,0)}. We showed that over the field of characteristic zero, if an open subvariety of a smooth affine surface is A^1-weakly equivalent to A^2-{(0,0)}, then it is isomorphic to A^2-{(0,0)}.
ABO blood-group gene frequencies in the Indian sub-continent: a statistical study of patterns of variation
(Indian Statistical Institute, Kolkata, 1980) Majumder, Partha P
Access structures for an image database
(Indian Statistical Institute, Kolkata, 1992) Kuila, Sudhansu Sekhar
Acyclicity Tests in Classes of Dense Digraphs in Streaming Model