Portfolio
MnM Hospital Multi-Agent Information Retrieval System
Multi-agent system helping healthcare workers quickly access institutional knowledge across hospital operations
On Translation Invariance in CNNs
Challenging the assumption of translation invariance in convolutional neural networks
Hallucination In Object Detection
Studying how object detectors can hallucinate non-existent parts with the DelftBikes dataset
Room-CanCell: AI-Driven Patient Room Automation
Simulated 3D hospital room that responds to patient vitals in real-time for pediatric cancer care
Joint Denoising and Classification for Edge Devices
Integrating image denoising and classification for resource-constrained devices using NAS
Aligning Bounding Boxes with Human Preference
Asymmetric loss function to align object detectors with human perception
Data-Efficient Visual Inductive Priors
Retrospective analysis of training deep models from scratch with limited data