OpenCV Machine Learning Note(I) kNN

Q1: What is kNN(k-Nearest Neighbor)?

from the document

• “kNN is one of the simplest classification algorithms available for supervised learning.”

Q2: What problem does KNN want to solve?

• “classification algorithms” is the key point

take one unclassified input data in to some predefined classes is a classification

Q3: What concepts in KNN?

• train data:

  because kNN need predefined pairs during the algorithms,
  the data making pairs is training data

• responses:

  responses are training data classified results, responses size should be equal to train data

• test data:

  a list of input data, each input data should be classified as one result of responses

• labels:

  the result of test data, same as responses, but with a different name

• distance:

  the number for judging neighbors, k nearest denote k minimums

Q4: What’s the main problem with kNN?

from document

"But there is a problem with that. Red Triangle may be the nearest. But what if there are a lot of Blue Squares near to him? "

if the number of red/blue are equal, the location of red/blue distribute maybe not, is it import?

1. how to choose the import k?
2. we are supposing all k neighbors are with equal importance? Is it justice?

赏

请问老板还招人么(/ω＼)

支付宝

微信