For binary classification (n = 2), we have only one mapping function f x = f+1 x −…
- User
For binary classification (n = 2), we have only one mapping function f x = f+1 x − f−1(x) as discriminant / scoring function. – The class labels are 𝒴 = −1, +1 . – The discrete class value is chosen by g x – g x ≔ f x ≥ δ with δ = 0.f(x) is referred to as confidence.= sgn f x or by thresholding• How many scoring functions do we have for a multi-class classification problem with 5 class categories?• For n-class classification (n ≥ 3), we have n discriminant / scoring functions f1, … , fn: 𝒳 → ℝ. – The class labels are 𝒴 = {1, … , n}.– The scores f1 x , … , fn(x) are transformed into discrete class values by choosing the class with the maximum score: g x = arg maxfi(x).
- Assistant
In a multi-class classification problem with ( n ) class categories (where ( n \geq 3 )), the concept of scoring functions expands from the binary case to accommodate multiple classes.
1. Scoring Functions for Multi-Class Classification
For ( n ) class categories, we need one discriminant or scoring function for each class. So, in the case of 5 class categories, ( n = 5 ). We denote the scoring functions as:
- ( f_1(x) ) for class 1
- ( f_2(x) ) for class 2
- ( f_3(x) ) for class 3
- ( f_4(x) ) for class 4
- ( f_5(x) ) for class 5
Thus, for 5 classes, the number of scoring functions we have is 5.
2. Class Label Assignment
The class labels are typically denoted as ( \mathcal{Y} = {1, 2, 3, 4, 5} ). The scoring functions ( f_1(x), f_2(x), ..., f_5(x) ) output real-valued scores for an input ( x ). The class label is assigned by taking the maximum score among all classes:
[ g(x) = \arg \max_{i} f_i(x) ]
This means that for a given input ( x ), the class with the highest score from the scoring functions is selected as the predicted class.
Summary
In conclusion, for a multi-class classification problem with 5 class categories, we have:
- 5 scoring functions: ( f_1(x), f_2(x), f_3(x), f_4(x), f_5(x) )
- Class assignments are determined by taking the maximum of these scoring functions.