Our paper G. Averkov, C. Hojny, M. Merkert: "On the Expressiveness of Rational ReLU Neural Networks With Bounded Depth" has been accepted at the International Conference on Learning Representations (ICLR) 2025 and selected for spotlight.
