[인공지능] 인공지능 개론

인공지능, Artificial Intelligence

🔎 인공지능?

Computer systems that can demonstrate human-level intelligence and replace human intellectual labout인공적으로, 즉 사람에 의해 만들어진 지능적인 시스템,사람의 지능적인 활동을 대신할 수 있는 컴퓨터 시스템이다.

 

🔎 머신러닝 Machine Learning ?

Computers learn from their past experience

Adapt to new environments or tasks

---

AI Agents

🔎AI Agents ?

Views of AI fall into four categories

	Humanly	Rationally : 주어진 정보기반 최적의 결론
Cognitive model	Thinking humanly	Thinking rationally
Active model	Acting humanly	Acting rationally (가장 많은 비중)

🔎 PEAS ?

Performance measure

Environment

Actuators

Sensors

 

• Example1 : Agent = taxi driver
  • Performance measure: Safe, fast, legal
  • Environment: Roads, traffic
  • Actuators: Steering wheel, accelerator
  • Sensors: Cameras, sonar, speedometer, GPS
• Example2 : Agent = Medical diagnosis
  • Performance measure: Healthy patient, minimize costs, lawsuits
  • Environment: Patient, hospital, staff
  • Actuators: Screen display (questions, tests, diagnoses, treatments, referrals)
  • Sensors: Keyboard (entry of symptoms, findings, patient's answers)

---

Environment types

1. Fully observable vs Partially observable
   • Fully observable: 전체 볼수 있나
   • Partially observable: 부분적으로 볼 수 있나
2. Deterministic vs Stochastic vs Strategic
   • Deterministic: 현재 상태에 따라 다음상태가 결정되는 경우
   • Strategic: agent 가 action을 취했을 때 다음 상태가 완전히 결정되는 경우
3. Episodic vs Sequential
   • Episodic: 다음에 영향을 미치지 않는 경우
4. Static vs Dynamic
   • Static: 변하지 않는 경우
   • Dynamic: 환경이 계속 변하는 경우
5. Single agent vs Multi agent

 

구분 중요하다고 함니다..

---

AI Systems: agents

 An agent is anything that can be viewed as perceiving its environment through sensors
and acting upon that environment through actuators

=> sensors 를 통해서 환경을 감지하고, actuators를 통해 action을 취한다.

 

🔎 Agents and environments

agent = architecture + program 

🔎 Agent types, Agent 종류

• Table Driven agents

 

• Simple reflex agents 
  • 과거 까지 다 보긴 힘드니까, 센서로 들어온 애만 판단

 

• Model-based reflex agents
  • 주어진 정보를 기반으로, 청소(action)를 하면 그 위치를 체크하는 등, 환경을 modeling 한다. 얼마나 변했나.

 

• Goal-based agents 
  • 모델링 + 목표

 

• Utility-based agents 
  • 모델링 + 목표 + 효율성

 

• Learning agent
  • 모델링 + 목표 + 효율성 + 경험