Skip to content
MeetupOnlinePast event

Advances in AI at Johns Hopkins University

Event artwork for Advances in AI at Johns Hopkins University

This event has ended

This listing is kept as an archive. Explore upcoming events to find your next one.

See upcoming events
About this event

What to expect

This virtual meetup brings together cutting-edge AI research from Johns Hopkins University, featuring three in-depth presentations from leading researchers pushing the boundaries of computer vision, medical imaging, and trustworthy AI systems. Whether you're a machine learning practitioner, computer vision enthusiast, or AI researcher, this session offers deep technical insights into state-of-the-art approaches solving real-world challenges in image generation, representation learning, and computational pathology.

Featured Presentations

Recent Advancements in Image Generation and Understanding by Aniket Roy, Research Scientist at NEC Labs America and recent PhD graduate from Johns Hopkins University. This presentation explores the evolution of image generation from Gaussian Mixture Models to diffusion models, with particular focus on three breakthrough works:

  • DiffNat (TMLR'25): A novel approach leveraging the kurtosis concentration property of natural images to produce sharper, more faithful generations across personalization, super-resolution, and unconditional synthesis tasks
  • DuoLoRA (ICCV'25): An efficient content-style personalization technique using adaptive-rank LoRA merging with cycle-consistency, achieving superior results with 19× fewer trainable parameters
  • Cap2Aug (WACV'25): A caption-based semantic augmentation strategy using cross-modal backtranslation to address data scarcity in few-shot and long-tail classification scenarios

From Representation Analysis to Data Refinement by Basudha Pal, recent PhD graduate from Johns Hopkins University's ECE Department. This talk examines how deep learning models encode unintended information and dependencies that affect reliability, fairness, and generalization. Dr. Pal introduces mutual information-based expressivity estimation to quantify implicit attribute encoding in learned embeddings, revealing hierarchical patterns across network layers. The presentation also covers data attribution techniques for identifying influential training samples that contribute to model errors, offering a unified data-centric perspective for analyzing and refining correlations in deep neural networks—particularly relevant for biometric and medical imaging applications.

Scalable & Precise Histopathology: Next-Gen Deep Learning for Digital Histopathology by [Tawsifur Rahman](https:

Running an AI event?

List it here, free.

Meetups, workshops and conferences go in front of thousands of builders after a quick review.

List your event