2020. 10. 16. 11:27ㆍ캐리의 데이터 세상/캐리의 데이터 공부 기록
코세라 딥러닝 문제 풀다가 틀린 내용들 관련 포스팅 모음
-
최적의 train / dev / test set split 관련 (스탠포드 CS230 블로그)
Splitting into train, dev and test sets
Splitting into train, dev and test sets
Splitting your data into training, dev and test sets can be disastrous if not done correctly. In this short tutorial, we will explain the best practices when splitting your dataset. This post follows part 3 of the class on “Structuring your Machine Learn
cs230.stanford.edu
- Weight decay
towardsdatascience.com/weight-decay-l2-regularization-90a9e17713cd
Weight Decay == L2 Regularization?
Neural Networks are great function approximators and feature extractors but sometimes their weights become too specialized and cause…
towardsdatascience.com
-
Inverted dropout technique
cs231n.github.io/neural-networks-2/#reg
CS231n Convolutional Neural Networks for Visual Recognition
Table of Contents: Setting up the data and the model In the previous section we introduced a model of a Neuron, which computes a dot product following a non-linearity, and Neural Networks that arrange neurons into layers. Together, these choices define the
cs231n.github.io
it is always preferable to use inverted dropout, which performs the scaling at train time, leaving the forward pass at test time untouched. Additionally, this has the appealing property that the prediction code can remain untouched when you decide to tweak where you apply dropout, or if at all.
'캐리의 데이터 세상 > 캐리의 데이터 공부 기록' 카테고리의 다른 글
| 코세라 - 딥러닝 테스트 기록 : Optimization algorithms (0) | 2020.11.04 |
|---|---|
| 10/29 공부기록 (2) | 2020.10.29 |
| 10/23 공부 기록 - 하이퍼파라미터 (0) | 2020.10.23 |
| 공부 기록 - django ~ (0) | 2020.10.14 |
| GitHub 신규 계정 Permission 오류 (0) | 2020.10.12 |
| 공개 데이터셋 리스트(참고자료) (0) | 2020.10.04 |
| Coursera - Deep Learning 수료증 / 코세라 실습 파일 다운받기 (2) | 2020.10.03 |