1

...Designing such a likelihood function is typically challenging; however, we observe that features like spectrogram are effective when

latent variables have limited degrees of freedom.This motivates us to infer latent variables via methods like Gibbs sampling, where we focus on approximating the conditional probability of a single variable given the others.

Above is an excerpt from a paper I've been reading, and I don't understand what the author means by degrees of freedom of latent variables. Could someone please explain with an example, or add more details?

### References

Shape and Material from Sound *(31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA)*

This seems to be correct, but maybe you should explain what would be the degrees of freedom of a latent variable in the specific context of that paper. – nbro – 2020-07-14T12:10:44.720

How does Gibbs sampling help in this context? – cogito_ai – 2020-07-15T14:10:31.697

1@cogito_ai Gibbs sampling helps us to approximate some observations from the distribution and compute the value of latent variables. – OmG – 2020-07-15T14:14:54.700