BERTVision
A novel architecture that treats BERT embeddings as visual tensors, achieving competitive NLP performance at a fraction of the computational cost through cross-domain transfer learning.
A novel architecture that treats BERT embeddings as visual tensors, achieving competitive NLP performance at a fraction of the computational cost through cross-domain transfer learning.
Statistical field experiment analyzing the impact of grammar quality on perceived intelligence in social media communications.
A distributed machine learning system built on Azure Databricks that predicts flight delays by processing large-scale aviation and weather datasets, enabling proactive operational optimization.
Transform PDFs into customizable multi-lingual podcasts with adjustable tone, length, and voice selection using GenAI technologies.
Instance segmentation challenge to detect cell nuclei in microscopy images for medical research. Top 12% placement in Kaggle competition.
Computer vision solution using neural network stacking to detect 15 facial landmarks, achieving 2nd place in Kaggle competition.
Statistical analysis of NC criminal data using linear regression models to provide evidence-based policy recommendations for crime reduction.
Data visualization and topic modeling analysis of 1.7M scholarly AI/ML articles from arXiv to identify research trends and industry patterns.
Multi-label, multi-class NLP classification challenge to identify topics in World Bank publications. 1st place finish with 0.6925 F1-micro score.
Regression analysis predicting college graduate earnings using U.S. Department of Education data. 1st place finish with 2.9796 RMSE.