Question: Question 2 ( Data Selection for Contrastive Learning ) Given the following data and latent classes, select a subset of size 4 using SAS. Compare

Question 2(Data Selection for Contrastive Learning)
Given the following data and latent classes, select a subset of size 4 using SAS.
Compare similarity between features using negative L1-Norm for simplicity. Thus, the optimal subset S** is defined here as:
S**=argminSkinK?iinSk?jinVk??Sk?||xi-xj||1
where k refers to the latent class of the examples
Vk refers to the set of examples in latent class k in the full data
Sk refers to the set of examples in latent class k in the selected subset
S=?kinKSk
Find S** i.e. the subset of size 4 that minimizes the above objective function.
K={0,1}
V0={2,3,4,5,7,9}
V1={1,6,8,10,11,12}
The examples are as follows:
x1=[0.0,0.0,1.5,3.0,-0.05], its latent class =1
x2=[4.0,6.0,0.0,0.0,0.05], its latent class =0
x3=[6.0,4.0,0.0,0.0,-0.05], its latent class =0
x4=[6.0,6.0,0.0,0.0,-0.05], its latent class =0
x5=[8.0,6.0,0.0,0.0,-0.05], its latent class =0
x6=[0.0,0.0,1.5,4.5,-0.05], its latent class =1
x7=[6.0,6.0,0.0,0.0,0.05], its latent class =0
x8=[0.0,0.0,1.5,6.0,0.05], its latent class =1
x9=[6.0,8.0,0.0,0.0,0.05], its latent class =0
x10=[0.0,0.0,1.0,4.5,0.05], its latent class =1
x11=[0.0,0.0,2.0,4.5,-0.05], its latent class =1
x12=[0.0,0.0,1.5,4.5,0.05], its latent class =1
Select indices of all examples that will be selected.
Question 2 ( Data Selection for Contrastive

Step by Step Solution

There are 3 Steps involved in it

1 Expert Approved Answer
Step: 1 Unlock blur-text-image
Question Has Been Solved by an Expert!

Get step-by-step solutions from verified subject matter experts

Step: 2 Unlock
Step: 3 Unlock

Students Have Also Explored These Related Programming Questions!