Papers of the day   All papers

Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation

Comments

Sep 11 2019 DeepMind

Neural networks in NLP are vulnerable to adversarially crafted inputs. We show that they can be trained to become certifiably robust against input perturbations such as typos and synonym substitution in text classification: https://arxiv.org/abs/1909.01492 https://t.co/bTrX55GWTG
8 replies, 694 likes


Sep 11 2019 pushmeet

Exciting new work from our team @DeepMindAI on development of AI systems for language understanding that are "provably" insensitive to even adversarial changes made to the input text.
0 replies, 10 likes


Sep 06 2019 arxiv

Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation. http://arxiv.org/abs/1909.01492 https://t.co/HCMgvX66mS
0 replies, 2 likes


Oct 11 2019 Benjamin Singleton

Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation #DataScience #BigData https://arxiv.org/abs/1909.01492
0 replies, 1 likes


Sep 06 2019 cs.LG Papers

Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation. Po-Sen Huang, Robert Stanforth, Johannes Welbl, Chris Dyer, Dani Yogatama, Sven Gowal, Krishnamurthy Dvijotham, and Pushmeet Kohli http://arxiv.org/abs/1909.01492
1 replies, 0 likes


Content