AI Gravity

Math 검색 결과

해당 글 3건

Leaner Regression (ft. Gradient descent, BackPropagation)

선형회귀 ----> 목적= "많은 데이터로 선긋기" $$y=ax+b\ \ \ \ \ \hat{y}=wx+b$$ \[ \begin{align} &\hat{y}는 예측값 \\ &w는 가중치 \\ &y는 타켓 \end{align} \] 1. w, b값 선정 w=1,b=1일때, \( \begin{align} \hat{y}&=wx+b \\&=1\cdot x\left[0\right]+1 \\ &\fallingdotseq 0.0619 \end{align} \) 2. w값 수정 (w 0.1 증가) \( \begin{align} \hat{y}' &=w'x+b \\&=\left(1+0.1\right)x\left[0\right]+1 \\ &\fallingdotseq 0.0675 \end{align} \) 3. Gradien..

Math/Concept 2021. 12. 2. 09:44

Loss Function List

회귀 타입에 쓰이는 손실함수는: MSE MAE MSLE MAPE KLD Poisson Logcosh Cosine Similarity Huber 분류에 쓰이는 손실함수는: Binary cross-entropy Categorical cross-entropy Sparse categorical cross-entropy Hinge Squared Hinge Categorical Hinge Segmentation에 쓰이는 손실함수: https://www.kaggle.com/code/bigironsphere/loss-function-library-keras-pytorch/notebook#Focal-Tversky-Loss Dice Loss BCE-Dice Loss Jaccard/Intersection over Union..

Math/Loss Function 2021. 11. 22. 16:07

SoftMax Function

SoftMax Function is make 100% ratio from elements. For example, input datas = { 1,2,3,4 } output datas = why we use e? 1. more clacification 2. easy to use differential import numpy as np import matplotlib.pyplot as plt def softmax(x): e_x = np.exp(x) return e_x / e_x.sum() x = np.array([1,2,3]) y = softmax(x) print(np.sum(y)) ratio = y labels = y print(y) print(ratio) print(labels) plt.pie(rati..

Math/Activation Function 2021. 11. 19. 10:48