Jensen Shannon Divergence || JS Divergence || Quick explained

in machine learning one of the fundamental task is to measure the similarity or dissimilarity between two probability distribution one common metric for this is the Jensen Channel Divergence which is a measure of how different two probability distribution are from each other in this video we'll explain what jst is how it's calculated and why it is important in machine learning GST is a statistical measure that used to compare two probability distributions it based on pullback labeler or care Divergence which I've already explained and you can watch it from here which is a measure of how different one probability distribution is from other but the KL Divergence is not symmetric means that the result will be different depending on depending on which distribution is considered as a reference distribution and which is a comparison distribution the GST is a symmetric version of KL Divergence that takes the average of KL Divergence between two distribution and their reverse this means the GST measures the similarity between two distribution regardless of which distribution is considered the reference to calculate the ghd we take the KL Divergence between the first distribution and the average of 2 distribution and then the KL Divergence between the second distribution and the average of 2 distribution this average is called Jensen Shannon Divergence GST is important in machine learning because it's often used as a loss function in generative models such as generative adversarial Network the model's aim to generate samples that resemble the original data distribution as closely as possible GSD is used as measure of the difference between generated samples and the original data distribution let's understand how it's calculated let's consider two probability distribution p and Q to calculate the JS Divergence we first need to calculate the average of two probability distribution we can do this by taking the element wise average of two distributions next we we need to calculate the KL Divergence between P and M and Q and M so by using the formula for KL Divergence we can calculate the Divergence between P and M and then we can calculate in the similar way for the Q and M finally we can calculate the JS Divergence by taking the average of KL Divergence between P and M and Q and M and hence the chair's Divergence between p and Q can be calculated like this therefore the GS Divergence between the two probability distribution p and Q is this in conclusion it's a symmetric version of KL Divergence that takes the average of Divergence between two distribution and their reverse GST is important in machine learning by understanding jsd you can better understand the principle Behind These models and how they are used in real world applications