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 |
- 분산 학습
- Time Series
- irregularly sampled time series
- doubleml
- causal machine learning
- causal reasoning
- 딥러닝
- 토픽모델링
- 리뷰
- causal transformer
- causal ml
- machine learning
- pytorch
- 의료
- 불규칙적 샘플링
- nccl 업그레이드
- Transformer
- GaN
- multi gpu
- NTMs
- causal forest
- 인과추론 의료
- 의료정보
- gru-d
- nccl 업데이트
- causal inference
- first pg on this rank that detected no heartbeat of its watchdog.
- 인과추론
- ERD
- netflix thumbnail
- Today
- Total
목록machine learning (2)
데알못정을

https://medium.com/@karanbhutani477/the-causal-revolution-in-machine-learning-moving-beyond-correlation-to-causation-07c4531c2cc0 The Causal Revolution in Machine Learning: Moving Beyond Correlation to CausationCausal Machine Learning (Causal ML) represents a fundamental shift in how we approach artificial intelligence, moving beyond simple pattern…medium.com위 블로그를 읽고 정리한 글임을 밝힙니다. 인과적 머신러닝의 이해인..

[regularization] 우리는 주어진 데이터를 가지고 어떤 예측을 수행할 때 모델을 만듭니다. 즉 prediction function 인데요, 이 함수가 우리가 가진 데이터를 너무 많이 fitting 할 경우엔 흔히 일반화 성능이 나오지 않는 과적합이 발생했다고 합니다. 따라서 모델이 학습하는 과정에서 어떤 규제를 주어 과적합을 방지 하겠다는 아이디어에서 나온 개념이 바로 ridge, lasso 규제 입니다. 우리의 모델이 선형모델인 경우에, 최소자승법(OLS)을 통해 loss 를 최소화 하는 조합의 weight를 구할 수 있는데, 여기 이 w^hat 이 바로 우리가 경험한 (empirical) 데이터에 대한 loss를 최소화하는 최적의 weight입니다. 규제를 걸어줄 경우에, min 값으로 묶..