GIST: Gauge-Invariant Spectral Transformers for Scalable Graph Neural Operators

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• GIST achieves end-to-end O(N) complexity through random projections while preserving gauge invariance.method 95%
• GIST achieves 99.50% micro-F1 on the PPI graph benchmark.result 90%
• GIST scales to mesh-based Neural Operator benchmarks with up to 750K nodes.technical 90%
• GIST achieves state-of-the-art aerodynamic prediction on DrivAerNet and DrivAerNet++ datasets.result 85%
• GIST enables discretization-invariant learning with bounded mismatch error.technical 90%
• Existing transformer positional encoding methods for graphs either require cubic-complexity eigendecomposition or break gauge invariance.limitation 85%
• GIST enables parameter transfer across arbitrary mesh resolutions for neural operator applications.technical 80%
• Failure to preserve gauge invariance causes catastrophic generalization failure in inductive learning.limitation 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card