CS231n 2강 요약 - Image Classification

미남로그·2021년 7월 12일
1

CS231n

목록 보기
1/8
post-thumbnail

| 0. 강의 자료

CS231n: Convolutional Neural Networks for Visual Recognition

| 1. 강의 목표

  1. K-Nearest Neighbor
  2. Linear classifiers: SVM, Softmax
  3. Two-layer neural network
  4. Image features

| 2. Image Classification

The Problem: Semantic Gap

An image is just a big grid of numbers between [0, 245]:

컴퓨터는 이미지를 0부터 245까지 픽셀로 본다. 따라서, 컴퓨터가 이미지 분류하기 위해서 몇 가지의 장애가 있다.

  1. Viewpoint variation
  2. Illumination
  3. Deformation (변형)
  4. Occulsion (가려짐)
  5. Background Clutter (배경과 유사)
  6. Intraclass variation (클래스 내부 분산)

이미지 접근 방법

  1. Collect a dataset of images and labels
  2. Use Machine Learning to train a classifier
  3. Evaluate the classifier on new images

First classifier: Nearest Neighbor

  • Memorize all data and labels
def train(images, labels):
	# Machine learning
    return model
  • Predic the label of the most similar training image
def predict(model, test_images):
	# Use model to predict labels
    return test_labels

| 3. K-Nearest Neighbors

Parameter of K-Nearest Neighbors

  1. K
  2. Distance Metric

| K closest points

Instead of copying label from nearest neighbor, take majority vote from K closest points

| Distance Metric(L1, L2)

| 4. Hyperparameter(Machine Learning)

In machine learning, a hyperparameter is a parameter whose value is used to control the learning process. By contrast, the values of other parameters (typically node weights) are derived via training.

These are hyperparameters: choices about the algorithm that we set rather than learn.

이미지 분류에선 잘 사용하지 않는 이유

  1. very slow at test time
  2. Distance metrics on pixels are not informative
  3. Curse of dimensionality

| 5. Linear Classification

In the field of machine learning, the goal of statistical classification is to use an object's characteristics to identify which class (or group) it belongs to. A linear classifier achieves this by making a classification decision based on the value of a linear combination of the characteristics. An object's characteristics are also known as feature values and are typically presented to the machine in a vector called a feature vector.

  • Hard cases for a linear classifier

profile
미남이 귀엽죠

0개의 댓글