Search
🌏

keras.1_b_2. title: 컴퓨터비전 contrastive learning 에 negative sample 이 최대한 많이 필요한 이유

🚀 prev note
♻️ prev note
🚀 next note
♻️ next note
17 more properties
컴퓨터비전 문제의 contrastive learning 에서 negative sample 이 엄청나게 많이 필요하다고 강조되는 이유는 이미지는 너무 고차원이고 연속적이기(참고1, 2) 때문이다. 이를 직관적으로 이해하기 위해서는 페이스북 블로그의 포스팅(참고 3)에 수록돼 있는 아래 그림(그림) 을 보면 좋다.
그림
•
그림
파란색 : compatible (x,y) pairs, 초록색 : incompatible (x,y) pairs. 각각을 positive, negative 라고 보아도 좋다.
참고
2.
Typical machine learning systems do so by treating the prediction problem as a classification problem and computing scores for each outcome using a giant so-called softmax layer, which transforms raw scores into a probability distribution over words. (설명1) With this technique, the uncertainty of the prediction is represented by a probability distribution over all possible outcomes, provided that there is a finite number of possible outcomes. (설명2) In CV, on the other hand predicting “missing” frames in a video missing patches in an image, or missing segment in a speech signal involves a prediction of high-dimensional continuous objects rather than discrete outcomes. (설명3) It is not possible to explicitly represent all the possible video frames and associate a prediction score to them. In fact, we may never have techniques to represent suitable probability distributions over high-dimensional continuous spaces, such as the set of all possible video frames. (설명4)