@article{bartolo-2022-models, author = {Max Bartolo and Tristan Thrush and Sebastian Riedel and Pontus Stenetorp and Robin Jia and Douwe Kiela}, title = {Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants}, journal = {CoRR}, volume = {abs/2112.09062}, year = {2021}, url = {https://arxiv.org/abs/2112.09062}, eprinttype = {arXiv}, eprint = {2112.09062}, timestamp = {Mon, 03 Jan 2022 15:45:35 +0100}, biburl = {https://dblp.org/rec/journals/corr/abs-2112-09062.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} }

@article{lu2021fantastically, title={Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity}, author={Lu, Yao and Bartolo, Max and Moore, Alastair and Riedel, Sebastian and Stenetorp, Pontus}, journal={arXiv preprint arXiv:2104.08786}, year={2021} }

@inproceedings{mozes-etal-2021-contrasting, title = "Contrasting Human- and Machine-Generated Word-Level Adversarial Examples for Text Classification", author = "Mozes, Maximilian and Bartolo, Max and Stenetorp, Pontus and Kleinberg, Bennett and Griffin, Lewis", booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing", month = nov, year = "2021", address = "Online and Punta Cana, Dominican Republic", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2021.emnlp-main.651", pages = "8258--8270", abstract = "Research shows that natural language processing models are generally considered to be vulnerable to adversarial attacks; but recent work has drawn attention to the issue of validating these adversarial inputs against certain criteria (e.g., the preservation of semantics and grammaticality). Enforcing constraints to uphold such criteria may render attacks unsuccessful, raising the question of whether valid attacks are actually feasible. In this work, we investigate this through the lens of human language ability. We report on crowdsourcing studies in which we task humans with iteratively modifying words in an input text, while receiving immediate model feedback, with the aim of causing a sentiment classification model to misclassify the example. Our findings suggest that humans are capable of generating a substantial amount of adversarial examples using semantics-preserving word substitutions. We analyze how human-generated adversarial examples compare to the recently proposed TextFooler, Genetic, BAE and SememePSO attack algorithms on the dimensions naturalness, preservation of sentiment, grammaticality and substitution rate. Our findings suggest that human-generated adversarial examples are not more able than the best algorithms to generate natural-reading, sentiment-preserving examples, though they do so by being much more computationally efficient.", }

@inproceedings{ niepert2021imle, author={Mathias Niepert and Pasquale Minervini and Luca Franceschi}, title={Implicit {MLE:} Backpropagating Through Discrete Exponential Family Distributions}, booktitle={Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS)}, year={2021}, url={https://arxiv.org/abs/2106.01798} }

@misc{bartolo2021improving, title={Improving Question Answering Model Robustness with Synthetic Adversarial Data Generation}, author={Max Bartolo and Tristan Thrush and Robin Jia and Sebastian Riedel and Pontus Stenetorp and Douwe Kiela}, year={2021}, eprint={2104.08678}, archivePrefix={arXiv}, primaryClass={cs.CL} }

@misc{lewis2021paq, title={PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them}, author={Patrick Lewis and Yuxiang Wu and Linqing Liu and Pasquale Minervini and Heinrich Küttler and Aleksandra Piktus and Pontus Stenetorp and Sebastian Riedel}, year={2021}, eprint={2102.07033}, archivePrefix={arXiv}, primaryClass={cs.CL} }