Neural Network Architecture Comparison Hub
3 papers - avg viability 4.0
Top Papers
- Rational ANOVA Networks(5.0)
Rational-ANOVA Networks improve interpretability and efficiency over traditional neural nets for better performance on compute-budgeted tasks.
- From Dead Neurons to Deep Approximators: Deep Bernstein Networks as a Provable Alternative to Residual Layers(4.0)
Deep Bernstein Networks optimize trainability and representation power without residual connections, offering enhanced expressive capacity over traditional architectures.
- Gradient Residual Connections(3.0)
Gradient-based residual connections improve neural network approximation of high-frequency functions, with potential application in image super-resolution.