Searcharxiv⌕ Search

arXiv subjects

Rohan Sharma

Publications and source records attributed to Rohan Sharma.

18 recordsLinked to original sources
arXiv2025

KidSpeak: A General Multi-purpose LLM for Kids' Speech Recognition and Screening

With the rapid advancement of conversational and diffusion-based AI, there is a growing adoption of AI in educational services, ranging from grading and assessment tools to personalized learning systems that provide targeted support for students. However, this adaptability has yet to fully extend to the domain of children's speech, where existing models often fail due to their reliance on datasets designed for clear, articulate adult speech. Children, particularly those in early developmental stages or with speech and language pathologies, present unique challenges that current AI models and datasets are ill-equipped to handle. To address this, we introduce KidSpeak, a multi-task speech-enhanced Foundation Model capable of both generative and discriminative tasks specifically tailored to children's speech patterns. Our framework employs a two-stage training process that incorporates phonetic knowledge into the speech encoder, achieving an average accuracy of 87% across four separate tasks. Furthermore, recognizing the limitations of scalable human annotation and existing speech alignment tools, we propose the Flexible and Automatic Speech Aligner (FASA) and leverage the method to construct high quality datasets for training and evaluation. This novel alignment tool significantly improves the quality of aligned children's speech from noisy data, enhancing data quality by 13.6x compared to human annotations, as demonstrated on the CHILDES dataset. To the best of our knowledge, KidSpeak and FASA represent the first comprehensive solution designed for speech and language therapy in children, offering both a multi-purpose speech LLM and a robust alignment tool.

eess.AS↗
arXiv2025

Inversion of Magnetotelluric Data using Bayesian Neural Networks

Magnetotelluric (MT) inversion is a key technique in geophysics for imaging deep subsurface resistivity structures. However, the inherent ill-posedness and non-uniqueness of inverse problems make them challenging to solve. While supervised deep learning approaches have shown promise in this domain, their predictions are typically deterministic and fail to capture the associated uncertainty, an essential factor for decision-making. To address this limitation, we explore the application of Bayesian Neural Networks (BNNs) for MT inversion with uncertainty quantification. Specifically, we train a Bayesian Convolutional Neural Network (BCNN) on a synthetically generated MT dataset. The BCNN effectively recovers resistivity profiles from apparent resistivity data, with the predicted means closely matching the ground truth across both the training and test sets, while also providing uncertainty estimates quantified within 3 standard deviations from the mean. These results underscore the potential of BNNs to enhance deep learning-based geophysical inversion frameworks by incorporating principled uncertainty quantification.

physics.geo-ph↗
arXiv2025

DETQUS: Decomposition-Enhanced Transformers for QUery-focused Summarization

Query-focused tabular summarization is an emerging task in table-to-text generation that synthesizes a summary response from tabular data based on user queries. Traditional transformer-based approaches face challenges due to token limitations and the complexity of reasoning over large tables. To address these challenges, we introduce DETQUS (Decomposition-Enhanced Transformers for QUery-focused Summarization), a system designed to improve summarization accuracy by leveraging tabular decomposition alongside a fine-tuned encoder-decoder model. DETQUS employs a large language model to selectively reduce table size, retaining only query-relevant columns while preserving essential information. This strategy enables more efficient processing of large tables and enhances summary quality. Our approach, equipped with table-based QA model Omnitab, achieves a ROUGE-L score of 0.4437, outperforming the previous state-of-the-art REFACTOR model (ROUGE-L: 0.422). These results highlight DETQUS as a scalable and effective solution for query-focused tabular summarization, offering a structured alternative to more complex architectures.

cs.CL↗
arXiv2025

Seismic inversion using hybrid quantum neural networks

Seismic inversion-including post-stack, pre-stack, and full waveform inversion is compute and memory-intensive. Recently, several approaches, including physics-informed machine learning, have been developed to address some of these limitations. Motivated by the potential of quantum computing, we report on our attempt to map one such classical physics-informed algorithm to a quantum framework. The primary goal is to investigate the technical challenges of this mapping, given that quantum algorithms rely on computing principles fundamentally different from those in classical computing. Quantum computers operate using qubits, which exploit superposition and entanglement, offering the potential to solve classically intractable problems. While current quantum hardware is limited, hybrid quantum-classical algorithms-particularly in quantum machine learning (QML)-demonstrate potential for near-term applications and can be readily simulated. We apply QML to subsurface imaging through the development of a hybrid quantum physics-informed neural network (HQ-PINN) for post-stack and pre-stack seismic inversion. The HQ-PINN architecture adopts an encoder-decoder structure: a hybrid quantum neural network encoder estimates P- and S-impedances from seismic data, while the decoder reconstructs seismic responses using geophysical relationships. Training is guided by minimizing the misfit between the input and reconstructed seismic traces. We systematically assess the impact of quantum layer design, differentiation strategies, and simulator backends on inversion performance. We demonstrate the efficacy of our approach through the inversion of both synthetic and the Sleipner field datasets. The HQ-PINN framework consistently yields accurate results, showcasing quantum computing's promise for geosciences and paving the way for future quantum-enhanced geophysical workflows.

quant-ph↗
arXiv2024

Enhancing Diffusion Posterior Sampling for Inverse Problems by Integrating Crafted Measurements

Diffusion models have emerged as a powerful foundation model for visual generations. With an appropriate sampling process, it can effectively serve as a generative prior for solving general inverse problems. Current posterior sampling-based methods take the measurement (i.e., degraded image sample) into the posterior sampling to infer the distribution of the target data (i.e., clean image sample). However, in this manner, we show that high-frequency information can be prematurely introduced during the early stages, which could induce larger posterior estimate errors during restoration sampling. To address this observation, we first reveal that forming the log-posterior gradient with the noisy measurement ( i.e., noisy measurement from a diffusion forward process) instead of the clean one can benefit the early posterior sampling. Consequently, we propose a novel diffusion posterior sampling method DPS-CM, which incorporates a Crafted Measurement (i.e., noisy measurement crafted by a reverse denoising process, rather than constructed from the diffusion forward process) to form the posterior estimate. This integration aims to mitigate the misalignment with the diffusion prior caused by cumulative posterior estimate errors. Experimental results demonstrate that our approach significantly improves the overall capacity to solve general and noisy inverse problems, such as Gaussian deblurring, super-resolution, inpainting, nonlinear deblurring, and tasks with Poisson noise, relative to existing approaches. Code is available at: https://github.com/sjz5202/DPS-CM.

cs.CV↗
arXiv2024

Real-Time Weapon Detection Using YOLOv8 for Enhanced Safety

This research paper presents the development of an AI model utilizing YOLOv8 for real-time weapon detection, aimed at enhancing safety in public spaces such as schools, airports, and public transportation systems. As incidents of violence continue to rise globally, there is an urgent need for effective surveillance technologies that can quickly identify potential threats. Our approach focuses on leveraging advanced deep learning techniques to create a highly accurate and efficient system capable of detecting weapons in real-time video streams. The model was trained on a comprehensive dataset containing thousands of images depicting various types of firearms and edged weapons, ensuring a robust learning process. We evaluated the model's performance using key metrics such as precision, recall, F1-score, and mean Average Precision (mAP) across multiple Intersection over Union (IoU) thresholds, revealing a significant capability to differentiate between weapon and non-weapon classes with minimal error. Furthermore, we assessed the system's operational efficiency, demonstrating that it can process frames at high speeds suitable for real-time applications. The findings indicate that our YOLOv8-based weapon detection model not only contributes to the existing body of knowledge in computer vision but also addresses critical societal needs for improved safety measures in vulnerable environments. By harnessing the power of artificial intelligence, this research lays the groundwork for developing practical solutions that can be deployed in security settings, ultimately enhancing the protective capabilities of law enforcement and public safety agencies.

cs.CV↗
arXiv2024

Understanding Fine-tuning in Approximate Unlearning: A Theoretical Perspective

Machine Unlearning has emerged as a significant area of research, focusing on `removing' specific subsets of data from a trained model. Fine-tuning (FT) methods have become one of the fundamental approaches for approximating unlearning, as they effectively retain model performance. However, it is consistently observed that naive FT methods struggle to forget the targeted data. In this paper, we present the first theoretical analysis of FT methods for machine unlearning within a linear regression framework, providing a deeper exploration of this phenomenon. Our analysis reveals that while FT models can achieve zero remaining loss, they fail to forget the forgetting data, as the pretrained model retains its influence and the fine-tuning process does not adequately mitigate it. To address this, we propose a novel Retention-Based Masking (RBM) strategy that constructs a weight saliency map based on the remaining dataset, unlike existing methods that focus on the forgetting dataset. Our theoretical analysis demonstrates that RBM not only significantly improves unlearning accuracy (UA) but also ensures higher retaining accuracy (RA) by preserving overlapping features shared between the forgetting and remaining datasets. Experiments on synthetic and real-world datasets validate our theoretical insights, showing that RBM outperforms existing masking approaches in balancing UA, RA, and disparity metrics.

cs.LG↗
arXiv2024

Joint Inversion of DC Resistivity and MT Data using Multi-Objective Grey Wolf Optimization

Joint inversion of geophysical datasets is instrumental in subsurface characterization and has garnered significant popularity, leveraging information from multiple geophysical methods. In this study, we implemented the joint inversion of DC resistivity with MT data using the Multi-Objective Grey Wolf Optimization (MOGWO) algorithm. As an extension of the widely-used Grey Wolf Optimization algorithm, MOGWO offers a suite of pareto optimal non-dominated solutions, eliminating the need for weighting parameters in the objective functions. This set of non-dominated predictions also facilitates the understanding of uncertainty in the predicted model parameters. Through a field case study in the region around Broken Hill in South Central Australia, the paper showcases MOGWO's capabilities in joint inversion, providing confident estimates of the model parameters (resistivity profiles), as indicated by a narrow spread in the suite of solutions. The obtained results are comparable to well established methodologies and highlight the efficacy of MOGWO as a reliable tool in geophysical exploration.

physics.geo-ph↗
arXiv2024

Inversion of DC Resistivity Data using Physics-Informed Neural Networks

The inversion of DC resistivity data is a widely employed method for near-surface characterization. Recently, deep learning-based inversion techniques have garnered significant attention due to their capability to elucidate intricate non-linear relationships between geophysical data and model parameters. Nevertheless, these methods face challenges such as limited training data availability and the generation of geologically inconsistent solutions. These concerns can be mitigated through the integration of a physics-informed approach. Moreover, the quantification of prediction uncertainty is crucial yet often overlooked in deep learning-based inversion methodologies. In this study, we utilized Convolutional Neural Networks (CNNs) based Physics-Informed Neural Networks (PINNs) to invert both synthetic and field Schlumberger sounding data while also estimating prediction uncertainty via Monte Carlo dropout. For both synthetic and field case studies, the median profile estimated by PINNs is comparable to the results from existing literature, while also providing uncertainty estimates. Therefore, PINNs demonstrate significant potential for broader applications in near-surface characterization.

physics.geo-ph↗
arXiv2024

Craft: Cross-modal Aligned Features Improve Robustness of Prompt Tuning

Prompt Tuning has emerged as a prominent research paradigm for adapting vision-language models to various downstream tasks. However, recent research indicates that prompt tuning methods often lead to overfitting due to limited training samples. In this paper, we propose a Cross-modal Aligned Feature Tuning (Craft) method to address this issue. Cross-modal alignment is conducted by first selecting anchors from the alternative domain and deriving relative representations of the embeddings for the selected anchors. Optimizing for a feature alignment loss over anchor-aligned text and image modalities creates a more unified text-image common space. Overfitting in prompt tuning also deteriorates model performance on out-of-distribution samples. To further improve the prompt model's robustness, we propose minimizing Maximum Mean Discrepancy (MMD) over the anchor-aligned feature spaces to mitigate domain shift. The experiment on four different prompt tuning structures consistently shows the improvement of our method, with increases of up to $6.1\%$ in the Base-to-Novel generalization task, $5.8\%$ in the group robustness task, and $2.7\%$ in the out-of-distribution tasks. The code will be available at https://github.com/Jingchensun/Craft

cs.CV↗
arXiv2024

Discriminative Adversarial Unlearning

We introduce a novel machine unlearning framework founded upon the established principles of the min-max optimization paradigm. We capitalize on the capabilities of strong Membership Inference Attacks (MIA) to facilitate the unlearning of specific samples from a trained model. We consider the scenario of two networks, the attacker $\mathbf{A}$ and the trained defender $\mathbf{D}$ pitted against each other in an adversarial objective, wherein the attacker aims at teasing out the information of the data to be unlearned in order to infer membership, and the defender unlearns to defend the network against the attack, whilst preserving its general performance. The algorithm can be trained end-to-end using backpropagation, following the well known iterative min-max approach in updating the attacker and the defender. We additionally incorporate a self-supervised objective effectively addressing the feature space discrepancies between the forget set and the validation set, enhancing unlearning performance. Our proposed algorithm closely approximates the ideal benchmark of retraining from scratch for both random sample forgetting and class-wise forgetting schemes on standard machine-unlearning datasets. Specifically, on the class unlearning scheme, the method demonstrates near-optimal performance and comprehensively overcomes known methods over the random sample forgetting scheme across all metrics and multiple network pruning strategies.

cs.LG↗
arXiv2023

DispersioNET: Joint Inversion of Rayleigh-Wave Multimode Phase Velocity Dispersion Curves using Convolutional Neural Networks

Rayleigh wave dispersion curves have been widely used in near-surface studies, and are primarily inverted for the shear wave (S-wave) velocity profiles. However, the inverse problem is ill-posed, non-unique and nonlinear. Here, we introduce DispersioNET, a deep learning model based on convolution neural networks (CNN) to perform the joint inversion of Rayleigh wave fundamental and higher order mode phase velocity dispersion curves. DispersioNET is trained and tested on both noise-free and noisy dispersion curve datasets and predicts S-wave velocity profiles that match closely with the true velocities. The architecture is agnostic to variations in S-wave velocity profiles such as increasing velocity with depth and intermediate low-velocity layers, while also ensuring that the output remains independent of the number of layers.

physics.geo-ph↗
arXiv2023

Integer Factorization through Func-QAOA

Integer factorization is a significant problem, with implications for the security of widely-used cryptographic schemes. No efficient classical algorithm for polynomial-time integer factorization has been found despite extensive research. Although Peter Shor's breakthrough quantum algorithm offers a viable solution, current limitations of noisy intermediate-scale quantum (NISQ) computers hinder its practical implementation. To address this, researchers have explored alternative methods for factorization suitable for NISQ devices. One such method is the Quantum Approximate Optimization Algorithm, which treats factoring as an optimization problem defined over binary bits, resulting in various problematic aspects. In this paper, we explore the Func-QAOA approach for factorization, which premises overcoming some of the limitations of previous approaches and allows the incorporation of more advanced factorization techniques. After reviewing the most promising quantum implementations for integer arithmetics, we present a few illustrative examples to demonstrate the efficacy of the Func-QAOA approach and discuss methods to reduce the search space to speed up the optimization process.

quant-ph↗
arXiv2023

Label-Retrieval-Augmented Diffusion Models for Learning from Noisy Labels

Learning from noisy labels is an important and long-standing problem in machine learning for real applications. One of the main research lines focuses on learning a label corrector to purify potential noisy labels. However, these methods typically rely on strict assumptions and are limited to certain types of label noise. In this paper, we reformulate the label-noise problem from a generative-model perspective, $\textit{i.e.}$, labels are generated by gradually refining an initial random guess. This new perspective immediately enables existing powerful diffusion models to seamlessly learn the stochastic generative process. Once the generative uncertainty is modeled, we can perform classification inference using maximum likelihood estimation of labels. To mitigate the impact of noisy labels, we propose the $\textbf{L}$abel-$\textbf{R}$etrieval-$\textbf{A}$ugmented (LRA) diffusion model, which leverages neighbor consistency to effectively construct pseudo-clean labels for diffusion training. Our model is flexible and general, allowing easy incorporation of different types of conditional information, $\textit{e.g.}$, use of pre-trained models, to further boost model performance. Extensive experiments are conducted for evaluation. Our model achieves new state-of-the-art (SOTA) results on all the standard real-world benchmark datasets. Remarkably, by incorporating conditional information from the powerful CLIP model, our method can boost the current SOTA accuracy by 10-20 absolute points in many cases.

cs.LG↗
arXiv2022

Clarinet: A Music Retrieval System

A MIDI based approach for music recognition is proposed and implemented in this paper. Our Clarinet music retrieval system is designed to search piano MIDI files with high recall and speed. We design a novel melody extraction algorithm that improves recall results by more than 10%. We also implement 3 algorithms for retrieval-two self designed (RSA Note and RSA Time), and a modified version of the Mongeau Sankoff Algorithm. Algorithms to achieve tempo and scale invariance are also discussed in this paper. The paper also contains detailed experimentation and benchmarks with four different metrics. Clarinet achieves recall scores of more than 94%.

cs.IR↗
arXiv2015

Self-Coordinated Corona Graphs: a model for complex networks

Recently, real world networks having constant/shrinking diameter along with power-law degree distribution are observed and investigated in literature. Taking an inspiration from these findings, we propose a deterministic complex network model, which we call Self-Coordinated Corona Graphs (SCCG), based on the corona product of graphs. As it has also been established that self coordination/organization of nodes gives rise to emergence of power law in degree distributions of several real networks, the networks in the proposed model are generated by the virtue of self coordination of nodes in corona graphs. Alike real networks, the SCCG inherit motifs which act as the seed graphs for the generation of SCCG. We also analytically prove that the power law exponent of SCCG is approximately $2$ and the diameter of SCCG produced by a class of motifs is constant. Finally, we compare different properties of the proposed model with that of the BA and Pseudofractal scale-free models for complex networks.

cs.DM↗
arXiv2015

Structural and Spectral properties of Corona Graphs

Product graphs have been gainfully used in literature to generate mathematical models of complex networks which inherit properties of real networks. Realizing the duplication phenomena imbibed in the definition of corona product of two graphs, we define Corona graphs. Given a small simple connected graph which we call seed graph, Corona graphs are defined by taking corona product of a seed graph iteratively. We show that the cumulative degree distribution of Corona graphs decay exponentially when the seed graph is regular and cumulative betweenness distribution follows power law when seed graph is a clique. We determine explicit formulae of eigenvalues, Laplacian eigenvalues and signless Laplacian eigenvalues of Corona graphs when the seed graph is regular. Computable expressions of eigenvalues and signless Laplacian eigenvalues of Corona graphs are also obtained when the seed graph is a star graph.

math.SP↗