Open Access. Powered by Scholars. Published by Universities.®

Digital Commons Network™

Open Access. Powered by Scholars. Published by Universities.®

Computer Sciences

Institution
Keyword
Publication Year
Publication
Publication Type
File Type

Articles 481 - 510 of 63009

Full-Text Articles in Entire DC Network

An Empirical Study Of Rlvr Fine-Tuning For Mathematical Problem Solving In Llms, Rashmi Konnur Jun 2026

An Empirical Study Of Rlvr Fine-Tuning For Mathematical Problem Solving In Llms, Rashmi Konnur

Master’s Dissertations

Large language models have shown immense improvement in coding and math performances thanks to reinforcement learning boosted algorithms. However, its true impact on broadening the reasoning and analytical capacities of an LLM is still contended. In this dissertation, we outline the foundations of Large Language Models, and delve into Reinforcement Learning with Verifiable Rewards (RLVR). We discuss various strategies to efficiently manipulate memory during a fine tuning update. We finally perform RLVR fine-tuning techniques on different models with varied use cases and compare their performances, which corroborate the efficiency of RLVR.

Go to article

Analysis And Machine Learning Adaptation Of A Cognitive Model For Human Memory, Trevor Cross, Aihua W. Wood Jun 2026

Analysis And Machine Learning Adaptation Of A Cognitive Model For Human Memory, Trevor Cross, Aihua W. Wood

Faculty Publications

In this paper, we use the Duolingo SLAM dataset to analyze several cognitive models of second language acquisition and develop new approaches for enhanced performance. In particular, we consider the Predictive Performance Equation and some of its underlying power laws. Leveraging insights from machine learning, we develop simple one-feature models as building blocks for combined models that match or in certain cases outperform the existing models at much reduced computational cost. In addition, a neural network with one fully connected hidden layer is constructed that outperforms all other models on sufficiently large datasets.

Go to article

Active Learning Of Constraint Boundaries Using Expected Magnitude Of Incorrectness And Neural Networks, Atticus Beachy, Ramana V. Grandhi Jun 2026

Active Learning Of Constraint Boundaries Using Expected Magnitude Of Incorrectness And Neural Networks, Atticus Beachy, Ramana V. Grandhi

Faculty Publications

This research proposes an acquisition function for constraint boundary identification, with applications to hypersonic air vehicles. Hypersonic vehicles endure extreme thermal loads caused by aerodynamic heating, resulting in a strong coupling between structural performance and aerothermodynamics. However, modeling coupled system behaviors requires simultaneous consideration of both aerodynamic and structural design variables, increasing the dimensionality of the design trade space and the difficulty of accurately modeling the constraints. Several active learning schemes have been proposed to accelerate identification of the composite feasible region that satisfies all constraints. Some of these require integrating the surrogate model over the entire design space with …

Go to article

Efficient Multimodal Foundation Model Tuning For Hallucination Mitigation, Fei Zhao Jun 2026

Efficient Multimodal Foundation Model Tuning For Hallucination Mitigation, Fei Zhao

ETDs from 2020-2029

Over the past few years, multimodal foundation models have achieved remarkable progress in perception and understanding. However, two challenges limit their reliability: (1) dependence on offline training, which in most real-world settings requires large volumes of labeled data and, as a result, hinders the model’s ability to adapt to new data or domains; (2) weak cross-modal grounding, which often leads to hallucinated content generation, producing descriptions that are linguistically fluent but inconsistent with the input visual evidence. This dissertation frames hallucination mitigation as an outcome of transitioning from fixed learning (static, offline fine-tuning) to adaptive, feedback-driven lifelong learning. By incorporating …

Go to article

Learnable Structured Attention For Student-Aware Knowledge Distillation In Dense Prediction, Chen Liu Jun 2026

Learnable Structured Attention For Student-Aware Knowledge Distillation In Dense Prediction, Chen Liu

ETDs from 2020-2029

Dense prediction tasks, including object detection and semantic segmentation, require models to produce structured predictions and are widely used in real-world vision applications. Although deep networks have achieved strong performance on these tasks, their high computational cost limits deployment on resource-constrained systems such as autonomous-driving vehicles. Knowledge distillation (KD) addresses this issue by transferring knowledge from a large teacher model to a compact student model. However, existing distillation methods for dense prediction face greater challenges than those for classification due to the more complex task requirements. To overcome the challenges, this thesis presents a unified study of adaptive distillation for …

Go to article

Development Of An Adaptive Pelican Crossing Model Using Fuzzy Logic In Mixed Traffic Conditions, Manazil Adam, Andyka Kusuma, R. Jachrizal Sumabrata Jun 2026

Development Of An Adaptive Pelican Crossing Model Using Fuzzy Logic In Mixed Traffic Conditions, Manazil Adam, Andyka Kusuma, R. Jachrizal Sumabrata

Smart City

Traffic management at at-grade pedestrian crossing facilities (pelican crossings) in highly populated areas, such as the Universitas Indonesia Station, faces significant inefficiency challenges. During peak hours, the fixed-time system is frequently disabled and replaced with subjective manual control by security personnel, thereby triggering irregular stop-and-go cycles and a high accumulation of vehicle delays. This study aims to develop a hybrid adaptive control model integrating Computer Vision, Genetic Algorithm (GA), and Fuzzy Logic to optimize intersection performance under mixed traffic conditions. The research methodology begins with the extraction of traffic and pedestrian characteristic data, calculated manually through recorded field observations. This …

Go to article

Spatial Tax Intelligence: Implementing Integrated Spatial Decision Support Systems For Smart Urban Governance, Wildan R. Irfana, Haldis A. Bahar, Muhammad H. Candra Jun 2026

Spatial Tax Intelligence: Implementing Integrated Spatial Decision Support Systems For Smart Urban Governance, Wildan R. Irfana, Haldis A. Bahar, Muhammad H. Candra

Smart City

The transition toward smart governance requires local governments to adopt data-driven policy approaches, particularly in managing urban tax capacity. However, tax administration in many developing regions remains predominantly tabular and administrative, limiting the ability to capture the spatial dynamics of economic activities. This reveals a critical knowledge gap, as integrated frameworks combining spatial and tax data to identify geographic disparities in tax performance remain limited. Consequently, areas with high economic activity but low tax compliance, referred to as tax blind spots, often remain undetected. This study aims to analyze the effectiveness of integrating spatial and tax data in identifying spatial-tax …

Go to article

Kernelizing Protein Interaction Languages: Spectral Approximations And Random Fourier Features, Aishik Ghosh Jun 2026

Kernelizing Protein Interaction Languages: Spectral Approximations And Random Fourier Features, Aishik Ghosh

Master’s Dissertations

Protein-peptide interactions play an important role in many biological phenomena, spanning adaptive immunity to disease pathology. In the Sliding Window Interaction Grammar (SWING) framework, interactions are represented as sequences of biochemical tokens embedded using Doc2Vec, allowing robust generalisation to unobserved MHC alleles. However, classification remains limited to a single Euclidean feature space that is incapable of resolving binding landscapes. This dissertation develops SWING for four distinct kernel types: Gaussian, Laplacian, anisotropic (ARD), and the Spectral Mixture (SM) kernel, each approximated using scalable Random Fourier Features. The SM kernel incorporates prior knowledge about secondary structure into its spectral density as biological …

Go to article

State Government, The Forgotten Cyber Actor, Joshua D. Strubel Jun 2026

State Government, The Forgotten Cyber Actor, Joshua D. Strubel

Doctoral Dissertations and Projects

Cyber incidents are among the most pervasive threats facing the United States, with the FBI recording over 859,000 reported attacks and an estimated $16.6 billion in losses in 2024 alone. Despite widespread recognition that effective cyber defense requires a whole-nation approach, the existing research literature overwhelmingly focuses on federal policy, leaving state governments as largely overlooked actors. This study addresses that gap by examining the research question: How does state cybersecurity policy affect malicious cyber actors' frequency of operations? Drawing on multilinear regression analysis augmented by Random Forest machine learning models, this study evaluates the relationship between state-level cyber deterrence …

Go to article

Learning In Infants Using Intrinsically Motivated Goal Conditioned Reinforcement Learning, T I Darsan Jun 2026

Learning In Infants Using Intrinsically Motivated Goal Conditioned Reinforcement Learning, T I Darsan

Master’s Dissertations

Traditional artificial intelligence models learn by passively digesting large datasets. In contrast, human infants discover skills by actively interacting with their bodies and environments without explicit external rewards. This thesis introduces the Composer Architecture, a machine learning framework designed to mimic this autonomous, open-ended development. The Composer architecture operates in a multi-stage loop, the latent model using Contrastive Learning Through Time (CLTT) to compress high-dimensional raw data from visual, proprioceptive, and touch sensors into a low-dimensional space. To preserve data relationships and prevent topological collapse, a Softmax activation forces these latent representations to lie smoothly on a probability simplex. A …

Go to article

Design And Evaluation Of A Code-Switching-Aware Multilingual Conversational Ai System Using Advanced Rag Architectures, Ashutosh Juvale Jun 2026

Design And Evaluation Of A Code-Switching-Aware Multilingual Conversational Ai System Using Advanced Rag Architectures, Ashutosh Juvale

Master’s Dissertations

Conversational artificial intelligence has become the primary interface through which hundreds of millions of users in India seek information and customer support. Yet the way these users actually write and speak is fundamentally at odds with the monolingual assumptions baked into most retrieval and generation systems: they code-switch, fluidly mixing one or more of the twenty-two scheduled languages of India with English, frequently typing Indic words in the Roman script ("mera refund kab tak aayega"). Standard Retrieval-Augmented Generation (RAG) pipelines silently fail on such input — the retriever returns off-topic passages because the query and the knowledge base live in …

Go to article

An Efficient Hierarchical Deployment Of Sensors For K-Coverage In Planner Wireless Sensor Network, Abhay Raj Singh Jun 2026

An Efficient Hierarchical Deployment Of Sensors For K-Coverage In Planner Wireless Sensor Network, Abhay Raj Singh

Master’s Dissertations

Ensuring reliable sensing coverage is a fundamental challenge in wireless sensor networks (WSNs), particularly when multiple sensors are monitoring each location to provide robustness against node failures. In this work, we address the problem of deterministic k-coverage in planar WSN by proposing a hierarchical triangular lattice-based deployment strategy that organizes sensor locations across di↵erent refinement levels and guarantees coverage of every point in the sensing domain by at least k sensors. Each lattice is three-colorable, and selective activation of color classes ensures adjustable coverage guarantees. We prove that activating a single color class at refinement level t guarantees at least …

Go to article

Telluric Correction Of M-Dwarf Stars Using Machine Learning, Sayak Rana Jun 2026

Telluric Correction Of M-Dwarf Stars Using Machine Learning, Sayak Rana

Master’s Dissertations

The study of M-dwarf stars is of prime scientific interest to us because of their closer habitable zones and the favorable conditions they offer for exoplanet detection. However, telluric contamination of the ground-based spectra results in sharp absorption lines, which makes their study cumbersome. Removing this contamination is necessary for estimating key stellar parameters. The central contribution is a one-dimensional Convolutional Neural Network (CNN) that retrieves the four atmospheric parameters governing telluric absorption: pressure, temperature, humidity, and airmass. These predicted parameters are passed to Telfit which produces an estimated telluric spectrum. The observed spectrum is then divided by this estimated …

Go to article

Get Ready To Lead: Human-Centered Leadership In An Ai-Driven World, Ellen Ramsey Jun 2026

Get Ready To Lead: Human-Centered Leadership In An Ai-Driven World, Ellen Ramsey

Faculty and Staff Publications & Presentations

As AI becomes increasingly integrated into our daily lives, online students continue to seek instructors who consistently appear, genuinely care about them as individuals, and provide guidance, challenge, and support. AI tools can help with speed and structure, but human-centered leadership keeps connection and meaning at the forefront of the learning experience. 

This interactive workshop invites online instructors and faculty leaders to explore how human-centered leadership can support their teaching in the middle of rapid technological change.

Grounded in a six-pillar leadership model that encompasses conscious self-awareness, relational intelligence, ethical influence, adaptive growth, transparent communication, and empowered action, this session …

Go to article

Tinyvgg-Based Real-Time Degradation Classification For Adverse Driving Scenes Using A Newly Collected Iraqi Driving Dataset, Yousif N. Abbas, Matheel E. Abdulmunim, Nada H. Ali, Ismail A. Mageed Jun 2026

Tinyvgg-Based Real-Time Degradation Classification For Adverse Driving Scenes Using A Newly Collected Iraqi Driving Dataset, Yousif N. Abbas, Matheel E. Abdulmunim, Nada H. Ali, Ismail A. Mageed

Journal of Soft Computing and Computer Applications

Environmental conditions such as low-light at night, fog scattering, glare artifacts, rain streaks, and rain smear distortions are significant issues of camera-based perception in Autonomous Vehicles (AVs). These degradations alter the statistics of the scene, mask structure, introduce non-uniform noise, and adversely affect downstream vision processes, including detection and tracking. To overcome this shortcoming, this paper presents a lightweight TinyVGG-based degradation classification system that runs in real time. The network extracts discriminative spatial features with hierarchical convolutional encoding and projects them to a lower-dimensional semantic representation with fully connected layers and a multi-class predictor based on SoftMax. In addition, a …

Go to article

Improving Approach Of Evolutionary Strategies For Clustering Technique Enhancement, Duaa Mahde Saleh, Hasanen S. Abdullah, Ahmad Zamsuri Jun 2026

Improving Approach Of Evolutionary Strategies For Clustering Technique Enhancement, Duaa Mahde Saleh, Hasanen S. Abdullah, Ahmad Zamsuri

Journal of Soft Computing and Computer Applications

The existence of the information has been the essential aspect of the whole society. Information is concentrated in all forms to be effectively utilized. Clustering — an unsupervised learning technique. It is based on data similarity that gives rise to issues in collection, challenges and instability in data structure. It proposes an advanced evolutionary method by combining two approaches. Firstly, it adopts the evolutionary approach and integrates the advantages between two methods to design one. Among them are Differential Evolution (DE) and Genetic Algorithm (GA), Evolutionary Strategy (ES) and Genetic Programming (GP), and Evolutionary Programming (EP) and Particle Swarm Optimization …

Go to article

A Comprehensive Review Of 1d Deep Learning Approaches In Facial Analysis: Face Recognition, Landmark Detection, And Mesh Modeling, Duaa J. Al Hammami, Rehab F. Hassan Jun 2026

A Comprehensive Review Of 1d Deep Learning Approaches In Facial Analysis: Face Recognition, Landmark Detection, And Mesh Modeling, Duaa J. Al Hammami, Rehab F. Hassan

Journal of Soft Computing and Computer Applications

Facial Analysis has progressed rapidly with deep learning and its 2D image-based models, especially Convolutional Neural Networks (CNNs), which have been the most popular methods. In recent years, 1D deep learning models have gained traction in the search for efficient solutions for face recognition, facial landmark detection, and 3D face mesh modeling. 1D models encode the facial structure as sequences, curves, or temporal signals, resulting in high computational efficiency, a small memory footprint, and good interpretability, making them well-suited for real-time and edge devices. This review is a step-by-step, organized exploration of 1D deep learning analysis of the face, its …

Go to article

Comparative Study On Throughput Optimization In Nfv: Traditional Dissemination Techniques Vs. Swarm Intelligence Approaches, Sanaa Salih Alwan, Asia Ali Salman, Wulfrano Arturo Luna Ramírez Jun 2026

Comparative Study On Throughput Optimization In Nfv: Traditional Dissemination Techniques Vs. Swarm Intelligence Approaches, Sanaa Salih Alwan, Asia Ali Salman, Wulfrano Arturo Luna Ramírez

Journal of Soft Computing and Computer Applications

Network Functions Virtualization (NFV) modernizes networks by replacing hardware with software, creating a more flexible network architecture and offering flexibility in dynamic network environments. This foundational technology is essential for creating the networks of the future, including the Internet of Things (IoT) and cellular services. NFV does provide flexibility, but it struggles to maintain system throughput during high traffic loads while achieving high resource utilization efficiency and dynamic packet routing. The problem lies in the fact that traditional request distribution mechanisms, such as flooding and gossip, fail to operate efficiently in complex network topologies (scale-free networks), leading to: (a) random …

Go to article

A Comprehensive Review Of The A* Algorithm: Evolution, Applications, And Future Trends In Path Planning, Saleel H. Abood, Hussein M. H. Al-Khafaji, Mohanned M. H. Al-Khafaji Jun 2026

A Comprehensive Review Of The A* Algorithm: Evolution, Applications, And Future Trends In Path Planning, Saleel H. Abood, Hussein M. H. Al-Khafaji, Mohanned M. H. Al-Khafaji

Journal of Soft Computing and Computer Applications

Despite being a fundamental problem to autonomous robotics and intelligent navigation systems, path planning is still a challenge. The A* algorithm is often used among search-based techniques for optimal search performance, as it's a tradeoff of computation. The above techniques have been developed for various applications as many versions of A* Dynamic A* (D*), D* Lite, Hybrid A*, and Anytime A* are suggested to deal with dynamic environments, real-time constraints, and kinematic restrictions. This paper comprehensively and structurally reviews the A* algorithm and its major extensions, encompassing historical development, methodological …

Go to article

Skin Lesion Classification Using Cnn Model And Augmented Dataset, Mohammed Nawzad Mohammed-Ramzi, Aso M. Aladdin Jun 2026

Skin Lesion Classification Using Cnn Model And Augmented Dataset, Mohammed Nawzad Mohammed-Ramzi, Aso M. Aladdin

Journal of Soft Computing and Computer Applications

Skin cancer is a deadly disease. Skin lesion classification is a critical challenge due to its prevalent and deadly nature. Skin lesions are difficult for dermatologists to detect using eye examination, which is time-consuming and variable. A deep learning model of skin lesions classification has been proposed using a Convolutional Neural Network (CNN) trained on the HAM10000 dataset of 10,015 dermatoscopies. To improve resilience and address the dataset's extreme class imbalance, data augmentation techniques such as geometric transformations, brightness/contrast adjustments, blurring, noise addition, histogram equalization, color space alterations, and elastic deformations are used. With a carefully balanced 10% test set, …

Go to article

Efficiency Improvement Of Rag Based Slm For Edge Devices, Pavan Prashanth Avanigadda Jun 2026

Efficiency Improvement Of Rag Based Slm For Edge Devices, Pavan Prashanth Avanigadda

Master’s Dissertations

The increasing need to deploy language models on constrained devices has given rise to efficiency issues in retrieval-augmented generation (RAG) approaches. Although RAGs boost answers’ quality by retrieving knowledge from external sources, current methods utilize static retrieval mechanisms, resulting in unnecessary computation, higher latencies, and inefficiency in resource usage. In this work, an efficient RAG approach based on small language models (SLMs) is presented, which uses a efficient and adaptive retrieval scheme. This method dynamically changes the retrieval depth and context constrution based on the complexity of the query, using a trained MLP router whose routing decisions are learned from …

Go to article

Developing A Model To Generate More Digital Data Of Indian Languages For Multilingual Applications, Arya Bagde Jun 2026

Developing A Model To Generate More Digital Data Of Indian Languages For Multilingual Applications, Arya Bagde

Master’s Dissertations

Most of India’s scheduled languages remain critically under-served by language technology because parallel (translated) text — the raw material that modern multilingual systems depend on — is extremely scarce. Back-translation can synthesise such data automatically, but its quality varies enormously, and unfiltered synthetic data can be worse than no data at all. This dissertation develops a framework that generates synthetic parallel data for four low-resource Indian languages spanning three language families and four scripts — Assamese (Indo-Aryan, Bengali script), Bodo (Tibeto-Burman, Devanagari), Manipuri (Tibeto-Burman, Bengali script) and Santali (Austroasiatic, Ol Chiki)—and introduces CASCADE, a learned multi-signal quality gate that scores …

Go to article

Dynamic Property Ordering For Efficient Multi-Property Bounded Model Checking, Vivek Kumar Jun 2026

Dynamic Property Ordering For Efficient Multi-Property Bounded Model Checking, Vivek Kumar

Master’s Dissertations

Formal verification plays a critical role in ensuring the correctness of modern hardware designs. As the complexity of digital systems increases, designs are often associated with a large number of verification properties that must be analyzed within limited computational resources. In conventional multi-property bounded model checking (BMC), all properties are verified simultaneously. While this approach enables parallel analysis, difficult properties can consume a disproportionate amount of resources, causing simpler properties to be delayed and reducing the overall efficiency of bug detection. This thesis presents dynamic property ordering techniques for efficient multi-property verification using SAT-based bounded model checking in the ABC …

Go to article

Cardiovascular Disease Subtypes And Alzheimer's Disease: Phenotypic And Genetic Associations In The Uk Biobank And All Of Us Research Program, Aili Toyli, Chen Zhao, Kuan Jui Su, Hui Shen, Hong Wen Deng, Qing Hui Chen, Qiuying Sha, Weihua Zhou Jun 2026

Cardiovascular Disease Subtypes And Alzheimer's Disease: Phenotypic And Genetic Associations In The Uk Biobank And All Of Us Research Program, Aili Toyli, Chen Zhao, Kuan Jui Su, Hui Shen, Hong Wen Deng, Qing Hui Chen, Qiuying Sha, Weihua Zhou

Michigan Tech Publications

BACKGROUND: Cardiovascular disease (CVD) and Alzheimer's disease (AD) are major public health concerns that share overlapping risk factors and potential mechanistic pathways. Although vascular contributions to cognitive decline are well documented, the specific relationships between AD and different CVD subtypes remain poorly understood. METHODS: In this cross-sectional study, we examined associations between AD and 11 CVD subtypes using logistic regression models in 2 large biobanks: the UK Biobank (n=502 133) and the All of Us Research Program (n=287 011). Models were adjusted for demographic, lifestyle, and clinical covariates. We also explored genetic overlap between AD and CVD traits through proximity-based …

Go to article

A System For The Prediction Of Election Results Using Vader And Hybridized Machine Learning Model, Abraham E. Evwiekpaefe, Khadijah Kabir, Georgina N. Obunadike Jun 2026

A System For The Prediction Of Election Results Using Vader And Hybridized Machine Learning Model, Abraham E. Evwiekpaefe, Khadijah Kabir, Georgina N. Obunadike

Tanzania Journal of Science

Integrating different classifiers along with sentiment lexicons like Vader, can enhance the performance of sentiment analysis systems. However, such a hybrid model remains underexplored, particularly in the context of regional elections in developing countries like Nigeria. The aim of this research is to develop a hybrid model that combines three machine learning classifiers and Vader lexicon to possibly achieve a higher accuracy. A case study of the 2023 governorship election in Kogi, Bayelsa and Imo State, Nigeria was examined. Twitter API library was utilized to extracted public and personal tweets using hashtags and keywords related to the target data from …

Go to article

Multi-Frequency Associative Memory For Continual Graph Learning Through Nested Optimization, Shuvam Kundu Jun 2026

Multi-Frequency Associative Memory For Continual Graph Learning Through Nested Optimization, Shuvam Kundu

Master’s Dissertations

Graph Neural Networks struggle to learn new tasks without forgetting old ones a problem known as catastrophic forgetting. In graph domains, this is compounded by structural shift, where newly added edges corrupt the learned representations of historical nodes even when model weights remain unchanged. We present CAM-Titans, a continual graph learning framework built around a two-buffer associative memory to address both parametric and structural forgetting. Our architecture operates across three timescales of adaptation: a slow base memory updated via ordinary gradient descent, an intermediate task buffer re-encoded after every task using the delta-rule, and a transient in-context state for rapid …

Go to article

A Study Of Prompt Tuning On Small Language Models(Slms): A Controlled Benchmark And A Lightweight Instance-Aware Method, Narkadamilli Sahith Jun 2026

A Study Of Prompt Tuning On Small Language Models(Slms): A Controlled Benchmark And A Lightweight Instance-Aware Method, Narkadamilli Sahith

Master’s Dissertations

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 …

Go to article

Reproducing And Analyzing The “Lost In The Middle” And “The Power Of Noise” Phenomenon In Retrieval-Augmented Generation, Kousik Samanta Jun 2026

Reproducing And Analyzing The “Lost In The Middle” And “The Power Of Noise” Phenomenon In Retrieval-Augmented Generation, Kousik Samanta

Master’s Dissertations

Retrieval-Augmented Generation has become the way to improve Large Language Models. They help with problems like knowledge and hallucinations. Recent studies show that these models still have limitations. One big problem is the “Lost in the Middle” phenomenon. Models can’t access information in the middle of contexts properly. Another counterintuitive observation is the “Power of Noise” paradigm, which suggests adding unrelated documents can actually make the generation better. We know these happen in extractive QA tasks, but we don’t know if they happen in tasks that need complex reasoning. This dissertation looks into how position and noise affect Long-Form Question …

Go to article

Learning To See Lesions, Not Skin Tone: Counterfactual Multimodal Learning For Fair, Trustworthy, And Text-Free Dermatology Ai, Shivam Jangid Jun 2026

Learning To See Lesions, Not Skin Tone: Counterfactual Multimodal Learning For Fair, Trustworthy, And Text-Free Dermatology Ai, Shivam Jangid

Master’s Dissertations

Recent advances in deep learning have significantly improved the performance of automated skin lesion classification systems, enabling accurate detection of various dermatological conditions from medical images. Despite these achievements, concerns regarding fairness and generalization remain a major challenge for the deployment of such systems in real-world clinical settings. A key factor contributing to these challenges is the presence of bias in training datasets, particularly with respect to skin-tone representation. Most publicly available skin lesion datasets contain a disproportionate number of samples from individuals with lighter skin tones. As a result, deep learning models trained on these datasets often learn representations …

Go to article

Adaptive Spectral Trust Gate For Physics- Constrained Operator Learning, Soham Chakraborty Jun 2026

Adaptive Spectral Trust Gate For Physics- Constrained Operator Learning, Soham Chakraborty

Master’s Dissertations

Physics-informed machine learning improves the plausibility, data-efficiency and generalization of surrogate models by injecting prior physical knowledge into the learning process. The current approaches can be broadly divided into two main categories: soft constraints, which add a physics residual to the training loss but guarantee nothing at inference time, and hard constraints, which project the model output onto the constraint set exactly but apply the projection uniformly to every part of the signal — including parts that are dominated by noise, discretization error, or model mismatch, where the idealized physics is not actually trustworthy. This dissertation proposes the Adaptive Spectral …

Go to article

First Prev Next Last