Adaptive Width Neural Networks accepted at ICLR 2026

Our paper Adaptive Width Neural Networks has been accepted for poster presentation at the International Conference on Learning Representations (ICLR) 2026, to be held at the Riocentro Convention and Event Center, Rio de Janeiro, Brazil, April 23–27, 2026. We introduce a method to learn the width of neural network layers during training (without upper bounds) and show easy truncation and dynamic compression. Co-authors: Federico Errica, Henrik Christiansen, Viktor Zaverkin, Mathias Niepert.