Adversarial ML Problems Are Getting Harder to Solve and to Evaluate
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929698321203200 |
|---|---|
| author | Rando, Javier Zhang, Jie Carlini, Nicholas Tramèr, Florian |
| author_facet | Rando, Javier Zhang, Jie Carlini, Nicholas Tramèr, Florian |
| contents | In the past decade, considerable research effort has been devoted to securing machine learning (ML) models that operate in adversarial settings. Yet, progress has been slow even for simple "toy" problems (e.g., robustness to small adversarial perturbations) and is often hindered by non-rigorous evaluations. Today, adversarial ML research has shifted towards studying larger, general-purpose language models. In this position paper, we argue that the situation is now even worse: in the era of LLMs, the field of adversarial ML studies problems that are (1) less clearly defined, (2) harder to solve, and (3) even more challenging to evaluate. As a result, we caution that yet another decade of work on adversarial ML may fail to produce meaningful progress. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_02260 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Adversarial ML Problems Are Getting Harder to Solve and to Evaluate Rando, Javier Zhang, Jie Carlini, Nicholas Tramèr, Florian Machine Learning Cryptography and Security In the past decade, considerable research effort has been devoted to securing machine learning (ML) models that operate in adversarial settings. Yet, progress has been slow even for simple "toy" problems (e.g., robustness to small adversarial perturbations) and is often hindered by non-rigorous evaluations. Today, adversarial ML research has shifted towards studying larger, general-purpose language models. In this position paper, we argue that the situation is now even worse: in the era of LLMs, the field of adversarial ML studies problems that are (1) less clearly defined, (2) harder to solve, and (3) even more challenging to evaluate. As a result, we caution that yet another decade of work on adversarial ML may fail to produce meaningful progress. |
| title | Adversarial ML Problems Are Getting Harder to Solve and to Evaluate |
| topic | Machine Learning Cryptography and Security |
| url | https://arxiv.org/abs/2502.02260 |