Bayesian CIM Paper Accepted for Oral Presentation at EPEPS 2026!

🎉 Thrilled to share that our paper “Bayesian Optimization of Crossbar-Based Compute-In-Memory System Design for Efficient DNN Inference” has been accepted for oral presentation at EPEPS 2026 in Fort Worth, Texas (Oct 18–21, 2026)!

We propose a multi-objective Bayesian optimization framework that co-optimizes hardware and algorithm parameters for crossbar-based CIM DNN accelerators, achieving substantial gains in area, energy, latency, and memory utilization on VGG8/CIFAR-10 and VGG16/Tiny-ImageNet-200 with comparable accuracy to baseline designs.

Authors: Arnob Saha, Bibhas Manna, Nikhil Kotikalapudi, Md Zesun Ahmed Mia, Rahul Kumar, Madhavan Swaminathan, Abhronil Sengupta

🔗 arXiv