Semi-Supervised Predictive Clustering Trees for (Hierarchical) Multi-Label Classification

Table 1

The proposed algorithm for learning of semi-supervised predictive clustering trees.

---

Procedure SSL-PCT

Input: A dataset , a parameter

Output: A predictive clustering tree

(1): 

(2): if then

(3):  for eachdo

(4):    = SSL-PCT(, )

(5):  return

(6): else

(7):  return

---

Procedure OptimizeParamW

Input: A dataset ; a set of values , ; a number of folds k

Output: A value

(1): for eachdo

(2):  

(3):   return

---

Procedure

Input: A dataset , a parameter

Output: The best test (), its heuristic score (), and the partition () it induces on the dataset (E)

(1): 

(2): for each possible test do

(3):  partition induced by on

(4):  

(5):  ifthen

(6):   

(7): return

---

Procedure CrossValidate

Input: A dataset , , a number of folds k

Output: An accuracy measure

(1):  = partition into k folds

(2): for eachdo

(3):   = SSL-PCT

(4):   = evaluate

(5): return

---