Mixquality_AL

Final Term Project For Bayesian Machine Learning Lecture (XAI-623)

Youtube Link

The presentation is given in YoutubeLink

Problem Formulation

Dirty MNIST

Active learning on MNIST: Ambiguous MNIST = 1:60

which is same with DDU paper.

OOD MNIST

Active learning on MNIST: Ambiguous MNIST : EMNIST = 1:60:1

Method

use MLN + Feature Density Traning process of MLN is same as paper and implementation. Estimation of feature density is done by modeling GMM with EM algorithm, implemented here

Expriment Results

Dirty MNIST

OOD MNIST