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 |
- 머신러닝
- SGD
- Vision
- 로지스틱 회귀
- 비용함수
- logistic regression
- SVM
- 파이썬
- CPP
- CNN
- C++
- OpenGL
- petal to metal
- pre-trained
- 신경망
- Kaggle
- 그래픽스
- Unsupervised learning
- 인공지능
- 컴퓨터 그래픽스
- neural network
- 백준
- 추천 시스템
- 컴퓨터 비전
- Support Vector Machine
- cs231n
- Computer Vision
- recommender system
- Regularization
- 딥러닝
- Today
- Total
목록딥러닝 (12)
kwan's note
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/VdShS/btqZOtp4aTD/A8EsVVbmuDMiiUfBxAyG81/img.png)
저번에는 google net 의 version 3인 inception v3를 이용해서 petal to metal의 classification을 진행하였습니다. 이번에는 조금 더 효율을 높이고자 augmentation을 하였고 또 inception v3모델이 아닌 최근 가장 강력한 모델중 하나인 efficient net을 이용하였습니다. www.kaggle.com/c/tpu-getting-started Petals to the Metal - Flower Classification on TPU Getting Started with TPUs on Kaggle! www.kaggle.com reminder-by-kwan.tistory.com/119 캐글 -petal to medtal (pre trained mod..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/cAt7W5/btqZQIGZYKx/hka9ZjFTbKcAKnmdDIR1dK/img.png)
www.kaggle.com/c/tpu-getting-started Petals to the Metal - Flower Classification on TPU Getting Started with TPUs on Kaggle! www.kaggle.com pedal to medal 을 오마주한 꽃 classification 이다. 104종류의 꽃을 classify하는 기본적인 image classification 문제인데 정말 간단한 방법으로 적당한 수준의 classification을 진행했다. 최고의 점수를 받기 위한 방법도 아니고 input도 전체중 일부만 사용하였다. input data도 다 사용하지 않고 상위30%의 점수를 받았으니 가성비는 좋다고 생각한다. def decode_image(image_d..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/bRMtIY/btqXrCtVPgs/QiyUUWntnmqKZctbAvLnkK/img.png)
출처: cs231n 이전 lecture reminder-by-kwan.tistory.com/116 linear classification with softmax - 소프트맥스 회귀 출처 : cs231n lecture3 www.youtube.com/watch?v=h7iBpEHGVNc 이번에는 저번시간에 이어서 linear classification의 방식에 대해 알아보도록 하겠습니다. reminder-by-kwan.tistory.com/115 linear classificatio.. reminder-by-kwan.tistory.com 지금까지 loss function을 구하고 gradient descent 방식을 통해 optimize를 진행했습니다. 이제는 neural network에서 중요한 개념중 하나..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/b23BFo/btqXj7Vo9uK/VTQVgvVxzcKrVV1bdlp3z0/img.png)
출처 : cs231n lecture3 www.youtube.com/watch?v=h7iBpEHGVNc 이번에는 저번시간에 이어서 linear classification의 방식에 대해 알아보도록 하겠습니다. reminder-by-kwan.tistory.com/115 linear classification, SVM - 선형 분류 출처 : cs231n lecture2 www.youtube.com/watch?v=OoUX-nOEjG0&list=PLC1qU-LWwrF64f4QKQT-Vg5Wr4qEE1Zxk&index=2 linear classification은 매우 기본이 되는 분류 방식이다. neural network의 근간이 되는 분류.. reminder-by-kwan.tistory.com 저번시간에 사용하였던..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/bwIRsG/btqWW9N8sg2/Mnq7zkmtTpRjcSRARD0421/img.png)
출처: cs231n lecture 2 www.youtube.com/watch?v=OoUX-nOEjG0&list=PLC1qU-LWwrF64f4QKQT-Vg5Wr4qEE1Zxk&index=2 이미지의 classification 문제중 가장 먼저 떠올릴수 있는 방법중 하나인 최근접이웃 알고리즘이다. 최근접 이웃 알고리즘은 말 그대로 가장 가까운 데이터의 결과값을 그대로 출력한다는것이다. 위와 같이 k=1 인경우 특정 데이터에 대해 점은 train set의 data고 색은 그 결과값(y)라고 할때 영역의 표시이다. 특정점에 대해 가장 가까운 점의 위치를 파악해 그 데이터의 결과값을 가져온다. 하지만 위처럼 섬이 생기게 되고 이는 옳지 못할 가능성이 매우 크다. 특이점에 매우 민감하게 반응하며 위와같은 실수를 범..
이 노트는 Fei-Fei li 교수님의 spring 2017 cs231 lecture를 기반으로 작성하였습니다. 작년말 이 강의를 추천받고 초반 3개정도의 강의를들었는데 당시에 머신러닝에 최소한의 개념도 부족해서 어느 순간부터 이해하지 못하고 포기했다. 이번 방학기간동안 몇개의 강의를 더 들었고 이제 다시 공부하고자 한다. CS231n: Convolutional Neural Networks for Visual Recognition www.youtube.com/watch?v=vT1JzLTH4G4&list=PLC1qU-LWwrF64f4QKQT-Vg5Wr4qEE1Zxk 이는 2017년 강의로 2020년강의 혹은 그 이후의 강의 슬라이드와 약간 변화가 있지만 전체 강의가 공개된 2017년강의를 기준으로 작성하였..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/wL9Vr/btqVFBEEOcy/u4vCpt62h59CLstFMMGud0/img.png)
출처: machine learning by andrew ng, stanford cousera lecture 수강일시: 2021.02.02 reminder-by-kwan.tistory.com/94 regularization - 정규화 출처: machine learning by andrew ng, stanford cousera lecture 수강일시: 2021.01.27 reminder-by-kwan.tistory.com/93 overfitting- 과적합 출처: machine learning by andrew ng, stanford cousera lecture.. reminder-by-kwan.tistory.com SVM의 수학적 배경이나 이해과정 이전에 먼저 타당성에 대한 검토를 하도록 하겠습니다. 이는..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/WkyF0/btqU4B5LNXi/149Awi91wO6mpkT0156XuK/img.png)
출처: machine learning by andrew ng, stanford cousera lecture 수강일시: 2021.01.28 reminder-by-kwan.tistory.com/96 neural network - 신경망 모델 출처: machine learning by andrew ng, stanford cousera lecture 수강일시: 2021.01.28 reminder-by-kwan.tistory.com/95 neural network -신경망 구조의 필요성 출처: machine learning by andrew ng, stanford.. reminder-by-kwan.tistory.com 이전까지는 신경망 모델이 어떻게 연구되기 시작했고 어떤 방식을 사용하는지 말했다면 이번시간에는 ..