Welcome to my digital workshop! I'm a final year B.Tech. Student in Data Science and AI at IIT Roorkee, I am passionate about 2D & 3D Computer Vision, Geometric Deep Learning, and Explainable & Adversarial ML.

• 🌐 Personal Website: shreesinghi.github.io
• 📧 Email: singhishree@gmail.com
• 📚 Google Scholar: Shree Singhi

• The Log-Diffusion Solver: Stabilizing Heat Diffusion on Discrete Meshes
  • MIT Summer Geometry Initiative (SGI) Research
  • Developed a novel PDE solver for heat diffusion on discrete mesh surfaces in the log-domain, providing enhanced numerical stability and including a formal proof of convergence.
• Mesh Fairing: Energy Minimization for Mesh Smoothing
  • MIT Summer Geometry Initiative (SGI) Research
  • Implemented and demonstrated surface fairing via energy minimization. Used a composite energy function combining Spring, Bending, and Angle Terms to effectively remove noise and improve triangle quality in crumpled 3D meshes.
• DINOHash: Learning Adversarially Robust Perceptual Hashes
  • ICML '25 CODEML Workshop
  • The first open-sourced, adversarially robust Deep Learning perceptual hashing model.
• Riemann Sum Optimization for Accurate Integrated Gradients Computation
  • NeurIPS '24 IAI Workshop
  • A novel framework that reduces computational costs for Integrated Gradient methods.
• Strengthening Interpretability: An Investigative Study of Integrated Gradient Methods
  • TMLR '24
  • Reproduced and extended the IDGI framework, correcting errors and conducting sensitivity analysis to identify conditions for improved attribution stability.