Erkut Erdem
2 papers · Latest:
Computer Vision
Beyond Gaussian Bottlenecks: Topologically Aligned Encoding of Vision-Transformer Feature Spaces
S$^2$VAE introduces a geometry-first VAE with hyperspherical latents, outperforming Gaussian bottlenecks in preserving 3D geometry and camera dynamics.
2604.28122
Artificial IntelligenceLearning to Think Like a Cartoon Captionist: Incongruity-Resolution Supervision for Multimodal Humor Understanding
This paper introduces IRS, a framework that uses incongruity-resolution supervision to teach multimodal models structured reasoning for humor understanding.
2604.15210
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.