Visualizing the Loss Landscape of Neural Nets.¶
Li, H., Xu, Z., Taylor, G., Studer, C., & Goldstein, T. (2018). Visualizing the Loss Landscape of Neural Nets. Advances in Neural Information Processing Systems.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Optimization Landscape
- The intervention is structural rather than algorithmic: add residual connections that smooth the landscape, apply layer normalisation that reshapes the curvature, or change the initialisation to place the model inside a basin — any of which changes the topology to one where the existing algorithm works.
This source2D slices show how skip connections and normalization smooth neural loss surfaces.
- The intervention is structural rather than algorithmic: add residual connections that smooth the landscape, apply layer normalisation that reshapes the curvature, or change the initialisation to place the model inside a basin — any of which changes the topology to one where the existing algorithm works.
Verification¶
This reference passed the adversarial substantiation pipeline: it was checked to exist and to support the claim it is attached to. See how references were verified.
Registry ID ref:d3fd614d5c1d · see in the full table