multiclass classification
-the one that called one-vs-all classification algorithm
classification with more than two groups
classification with more than two groups
either starting from index 0 or 1 doesn't matter..
the defined algorithms earlier is to compute the binary by using logistic regression...
using one vs all algorithm for solving multiclass classification.
- the algorithm is essentially make two groups, with each class again the rest...and take each hypothesis(that denotes by superscript) for each class
- then every h is trained to predict the probability
- the i which has the highest probability (highest confident that predict y) will picked as the basis classifier.
- so for every x, whichever most believe (most confident) will get y = i for the corresponding x
- so this algorithm is benefit for multiclass classification