arXiv subjects

Yash Singh

Publications and source records attributed to Yash Singh.

3 recordsLinked to original sources

arXiv2026

Beyond WER: Entity and Disfluency Recall in Accented Conversational ASR

ASR systems optimised for Word Error Rate (WER) often miss named entities and filled pauses in accented conversational English, both critical for language-learning feedback. We present a three-stage pipeline for speakers from India, Indonesia, and Latin America: (1) heuristic SQL filters curating entity-rich training data at 2.8x the entity density of random sampling, (2) regional LoRA adapters fine-tuned on Qwen2.5-Omni-3B producing both verbatim and corrected transcripts in a single forward pass, and (3) a six-category error taxonomy validated by an LLM-based judge (83.8% agreement, 210 human-labelled samples). The pipeline achieves 80-85% entity recall (up from 53-55%), 76-86% filler recall (up from <5%), and 6-10% WER across 6k test utterances, outperforming Whisper and a commercial ASR on entity recall while matching a zero-shot 30B model with 10x fewer parameters. Paired bootstrap tests confirm that curation alone accounts for 2.8-4.2 pp of entity recall gain (p<0.0001).

cs.CL↗

arXiv2024

A Multi-agent Simulation for the Mass School Shootings

The increasing frequency of mass school shootings in the United States has been raised as a critical concern. Active shooters kill innocent students and educators in schools. These tragic events highlight the urgent need for effective strategies to minimize casualties. This study aims to address the challenge of simulating and assessing potential mitigation measures by developing a multi-agent simulation model. Our model is designed to estimate casualty rates and evacuation efficiency during active shooter scenarios within school buildings. The simulation evaluates the impact of a gun detection system on safety outcomes. By simulating school shooting incidents with and without this system, we observe a significant improvement in evacuation rates, which increased from 16.6% to 66.6%. Furthermore, the Gun Detection System reduced the average casualty rate from 24.0% to 12.2% within a period of six minutes, based on a simulated environment with 100 students. We conducted a total of 48 simulations across three different floor layouts, varying the number of students and time intervals to assess the system's adaptability. We anticipate that the research will provide a starting point for demonstrating that a gunshot detection system can significantly improve both evacuation rates and casualty reduction.

cs.CY↗

arXiv2024

Persistence Image from 3D Medical Image: Superpixel and Optimized Gaussian Coefficient

Topological data analysis (TDA) uncovers crucial properties of objects in medical imaging. Methods based on persistent homology have demonstrated their advantages in capturing topological features that traditional deep learning methods cannot detect in both radiology and pathology. However, previous research primarily focused on 2D image analysis, neglecting the comprehensive 3D context. In this paper, we propose an innovative 3D TDA approach that incorporates the concept of superpixels to transform 3D medical image features into point cloud data. By Utilizing Optimized Gaussian Coefficient, the proposed 3D TDA method, for the first time, efficiently generate holistic Persistence Images for 3D volumetric data. Our 3D TDA method exhibits superior performance on the MedMNist3D dataset when compared to other traditional methods, showcasing its potential effectiveness in modeling 3D persistent homology-based topological analysis when it comes to classification tasks. The source code is publicly available at https://github.com/hrlblab/TopologicalDataAnalysis3D.

cs.CV↗