EPFL Papers @ ICLR 2021

© 2021 EPFL

© 2021 EPFL

EPFL researchers will be presenting the following 7 papers at ICLR 2021, the International Conference on Learning Representations.
Due to Covid-19 measures this year edition will be held virtually from 4th to 8th May 2021.
If you are attending ICLR 2021, we hope to meet and discuss!

List of Accepted EPFL Papers:

  • Distributed Momentum for Byzantine-Resilient Stochastic Gradient Descent; authors: E-M. El-Mhamdi, R. Guerraoui and S. Rouault
  • Rethinking the Role of Gradient-based Attribution Methods for Model Interpretability; authors: S. Srinivas and F. Fleuret
  • Taming GANs with Lookahead-Minmax; authors: T. Chavdarova, M. Pagliardini, S. Stich, F. Fleuret, and M. Jaggi
  • Group Equivariant Stand-Alone Self-Attention For Vision; authors: by D. W. Romero and J-B. Cordonnier
  • Understanding the Effects of Data Parallelism and Sparsity on Neural Network Training; authors: N. Lee, T. Ajanthan, P. Torr and M. Jaggi
  • Variational Information Bottleneck for Effective Low-Resource Fine-Tuning; authors: R. K. Mahabadi, Y. Belinkov and J. Henderson
  • On the Stability of Fine-tuning BERT: Misconceptions, Explanations, and Strong Baselines; authors: M. Mosbach, M. Andriushchenko, D. Klakow