AI RESEARCH
Slithering Through Gaps: Capturing Discrete Isolated Modes via Logistic Bridging
arXiv CS.LG
•
ArXi:2604.10821v1 Announce Type: new High-dimensional and complex discrete distributions often exhibit multimodal behavior due to inherent discontinuities, posing significant challenges for sampling. Gradient-based discrete samplers, while effective, frequently become trapped in local modes when confronted with rugged or disconnected energy landscapes. This limits their ability to achieve adequate mixing and convergence in high-dimensional multimodal discrete spaces.