Skip to content

Angledsugar/MAPA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MAPA — Multi-Agent Poisoning Attacks

Research line on reward-poisoning attacks against multi-agent reinforcement learning.


Current direction (main): Reward-Poisoning Carryover

Heal or Amplify? Reward-Poisoning Carryover from Single-Agent to Cooperative Multi-Agent Locomotion

A single-agent locomotion policy is trained with a goal-reaching reward (reach box → reward) and poisoned via reward poisoning (목표를 decoy로 redirect). That poisoned policy is then transferred/extended into a cooperative, shared-reward multi-agent setting. Central question:

shared-reward 협조 재학습은 transferred poison을 치유(heal) 하는가, 아니면 credit-assignment를 타고 clean agent에게 전염·증폭(amplify) 하는가?

→ Full research design: docs/research-design.md

Key design decisions (확정)

  • 이질 오염 — N개 중 k개 agent만 poisoned, 나머지 clean (진짜 MARL spillover 관측).
  • 두 전이 방식 비교 — zero-shot 배포 ∧ warm-start + MARL 재학습.
  • coordination 결합 — 공동 task(coverage/formation) → 피해가 선형 아닌 cliff.
  • money plot — ④(retrain,poison) − ②(zero-shot,poison) 의 부호.

Repository layout

경로 내용
docs/ research design, notes
experiments/ 실험 정의 — 2×2 factorial(init × transfer-mode) + sweeps(k, poison 깊이 ε)
configs/ environment / poison / training 설정
src/ 구현 (stack TBD)
scripts/ entry points — poison → transfer → evaluate → plot

Status

  • Research design — docs/research-design.md
  • Stack 결정 — IsaacLab 쿼드러페드(design doc 매칭) vs Unity ML-Agents(MAPA 확장)
  • Environment — multi-goal coverage / formation (loosely-coupled coordination; §6 floor-confound 회피)
  • Stage 1 — single-agent reward poisoning (decoy redirect, depth ε sweep)
  • Stage 2 — transfer (zero-shot ∧ warm-start+retrain), k sweep
  • Analysis — heal-vs-amplify money plot, spillover 채널 분리

History

  • Humanoids 2025 (Late-Breaking Report)Poisoning Attacks on Multi-Agent RL Systems (Choi, Cho, Lee). Unity ML-Agents Crawler, attacker-in-environment poison-cube paradigm; PPO/SAC × single/multi crawler. 전체 코드·빌드는 브런치 humanoids2025-lbr 에 보존 (origin push는 검토 후).

⚠️ 현재 방향은 LBR의 단순 연장이 아니라 위협 모델 피벗: attacker-in-env(online cube placement)pre-poisoned policy carryover(checkpoint 재사용/transfer).

About

Multi-Agent Deep Reinforcement Learning to survive an Poisoning Attack

Resources

License

Stars

3 stars

Watchers

1 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors