0123

<RNN>

• 시계열 데이터의 시간 속성을 충분히 이용하기 위해
• RNN 학습은 BPTT로

• Vanishing gradient problem 발생
• Sol : RNN Variants -> 1,2,3

1. LSTM

 

(*cell : 장기기억)

 

*gate 4가지

 

 

(1) forget 

 

 

(2) input

 

 

(3) output

 

 

---

 

2. GRU

 

->LSTM과 비슷

 

 

(1) update

 

 

(2) reset

 

 

(3) memory candidate

 

(4) final memory

---

 

3. ReLU

 

->생각보다 (-)

 

---

RNN Models.pdf

9.86MB

RNN_LSTM_GRU(최종).pdf

3.82MB

 

 

 

---

https://drive.google.com/file/d/1FzOrIIOPr2AM0klKj6ig9C0--ajtC3a8/view?usp=sharing

Image Segmentation with GrabCut Algorithm.ipynb

https://drive.google.com/file/d/1GDiLbPsoMguuiqVhFw0lW_Lx-dn2o0p3/view?usp=sharing

Image Segmentation with Watershed.ipynb

https://drive.google.com/file/d/1FyMRHzMsKBERiQ7H7k-cL9IRPSV2bj0o/view?usp=sharing

OpenCV영상처리_프로젝트.ipynb