Semi-supervised Learning

1. What is Semi-supervised Learning

1. Supervised Learning

   labeled data:\(\{(x^r,\hat{y}^r\}_{r=1}^R\)

   E.g: image,\(\hat{y}^r\): class labeles

2. Unsupervised Learning

   unlabeled data:\(\{x^r\}_{r=1}^R\)

   E.g: Clustering problem

3. Semi-supervised Learning

   both labeled data and unlabeled data:\(\{(x^r,\hat{y}^r)\}_{r=1}^R,\{x^u\}_{u=R}^{R+U}\)

   • A set of unlabeled data,usually U >> R
   • Transductive Learning: unlabeled data is the testing data
   • Inductive Learning: unlabeled data is not the testing data

2. Why Semi-supervied Learning

• It's easy to collect unlabeled data
• We do semi-supervied learning in our lives