일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 | 31 |
- Generative model
- Transfer Learning
- SOMs
- VGGNet
- MLOps
- AI 윤리
- 경사하강법
- tensorflow
- Python
- ResNet
- textmining
- Attention
- gaze estimation
- 군집화
- stemming
- Support Vector Machine
- NER
- Logistic Regression
- Ann
- NMF
- BERT
- 자기조직화지도
- LSTM
- cross domain
- TFX
- Clustering
- RNN
- Gradient Descent
- nlp
- Binary classification
- Today
- Total
juooo1117
[Module 5] 인과추론: Causality 본문
Part 1. Causality
What is Causality?
Causality is influence by which one event, process, state, or object (a cause) contributes to the production of another event, process, state, or object (an effect) where the cause is partly responsible for the effect, and the effect is partly dependent on the cause.
→ 하나의 어떤 무엇인가가 다른 무엇을 생성함에 있어서 영향을 미치는 것! (인과성)
기계학습 : 데이터의 상관성 학습
환경에 변화를 줌으로써 원하는 최종적인 상태로 변화시켜 나가는 인과관계로 해석할 수 있다.
Lesson’s Learned from Simpson’s Paradox
무작위 상태에서 인과관계를 추론할 수 있음!
- Causal analyses need to be guided by subject-matter knowledge.
- Identical data arising from different causal structures need to be analysed differently.
- No purely statistical rules exist to guide causal analyses.
데이터를 분석하고자 할 때 고려해야하는 두 가지!
Data scientists should take care of the types of data and question:
→ 주어진 데이터가 상관성을 지니고 있는지 아니면 인과성을 지니고 있는지를 확인해야 한다. 그리고 우리가 알고자 하는 질문이 단순히 조건부 확률같은 상관성에 관한 것인지, 인과성에 관한 것인지를 확인해야 한다.
Causal Inference (인과추론)
우리가 알 수 없는 실험 결과를 관측 데이터와 연결하는 것 → 두 가지를 연결하려면 블랙박스에 대한 형식적, 수학적인 이해가 필요하다.
Causal Framework: Structural Causal Model
SCM is a formal framework to study causality
모든 관측가능한 변수들의 값을 생성해내는 인과적 메커니즘을 정의하고 있음!
- 이 모델을 기반으로 우리가 어떤 중재도 하지 않는다면, 관측 가능한 모든 변수들에 대한 관측 분포를 볼 수 있다.
- 중재를 한다면, 해당 실험에 대한 결과 분포가 나올 것이다.
Structural Causal Models
Intervention (중재)
변수 x를 중재한다고 하는 것은?
: Given a model M the action of fixing any observable variable X ∈ V to a constant value x is denoted using the do(·) operator as do(X = x).
This operation gives birth to a submodel Mx that looks exactly like M but with functions where fx has been replaced with a constant x.
These two graphs represent the world before and after an intervention do(X = x).
Causal Effects (인과효과)
임의의 변수 집합 X가 있고, 이 집합 X가 고정되었을 때 관심을 가지고 보는 변수 Y 들이 특정값을 가지는 확률은 어떻게 될까를 보는 것.
(Given two disjoint sets of variables, X and Y, the causal effect of X on Y, denoted as P(y|do(x)) or Px(y), is a function from X to the space of probability distributions of Y.)
'Artificial Intelligence > LG Aimers: AI전문가과정' 카테고리의 다른 글
[Module 8] Recommender System (0) | 2024.01.19 |
---|---|
[Module 8] B2B 고객 행동 예측 방법론 (0) | 2024.01.19 |
[Module 6] Deep Learning: Self-Supervised Learning & Pre-Trained Models (0) | 2024.01.14 |
[Module 6] Deep Learning: Transformer (1) | 2024.01.14 |
[Module 6] Deep Learning: Seq2Seq with Attention for Natural Language Understanding and Generation (0) | 2024.01.13 |