MétaCan
Menu
Back to cohort
Record W4391903608 · doi:10.24251/hicss.2023.799

Safe Reinforcement Learning via Observation Shielding

2023· article· en· W4391903608 on OpenAlex
Joe McCalmon, Tongtong Liu, Andrew Cyhaniuk, Talal Halabi, Sarra Alqahtani

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenueProceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences · 2023
Typearticle
Languageen
FieldComputer Science
TopicAdversarial Robustness in Machine Learning
Canadian institutionsUniversité Laval
FundersNational Science Foundation
KeywordsRobustness (evolution)Computer scienceAdversarial systemAgile software developmentSAFERArtificial intelligenceMachine learningReinforcement learningDeep neural networksArtificial neural networkComputer security

Abstract

fetched live from OpenAlex

Reinforcement Learning (RL) algorithms have shown success in scaling up to large problems. However, deploying those algorithms in real-world applications remains challenging due to their vulnerability to adversarial perturbations. Existing RL robustness methods against adversarial attacks are weak to large perturbations - a scenario that cannot be ruled out for RL adversarial threats, as is the case for deep neural networks in classification tasks. This paper proposes a method called observation-shielding RL (OSRL) to increase the robustness of RL against large perturbations using predictive models and threat detection. Instead of changing the RL algorithms with robustness regularization or retrain them with adversarial perturbations, we depart considerably from previous approaches and develop an add-on safety feature for existing RL algorithms during runtime. OSRL builds on the idea of model predictive shielding, where an observation predictive model is used to override the perturbed observations as needed to ensure safety. Extensive experiments on various MuJoCo environments (Ant, Hooper) and the classical pendulum environment demonstrate that our proposed OSRL is safer and more efficient than state-of-the-art robustness methods under large perturbations.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.006
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow), Science and technology studies, Scholarly communication, Open science
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Theoretical or conceptual · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.967
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0060.000
Meta-epidemiology (narrow)0.0010.001
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.004
Science and technology studies0.0020.002
Scholarly communication0.0020.005
Open science0.0150.003
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.054
GPT teacher head0.304
Teacher spread0.250 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it