Wide Two-Layer Networks can Learn from Adversarial Perturbations
–Neural Information Processing Systems
In this study, we theoretically explain the counterintuitive success of perturbation learning. We assume wide two-layer networks and the results hold for any data distribution.
Neural Information Processing Systems
Feb-15-2026, 16:32:53 GMT
- Country:
- Asia
- Japan > Honshū
- Kantō > Tokyo Metropolis Prefecture > Tokyo (0.04)
- Middle East > Jordan (0.04)
- Japan > Honshū
- North America > Canada
- Asia
- Genre:
- Research Report
- Experimental Study (0.92)
- New Finding (1.00)
- Research Report
- Industry:
- Information Technology (0.45)
- Technology: