1

I want to understand the structure of a fully connected Boltzmann machine. I know that it is a generative network and that it has visible and hidden nodes. Do I understand correctly that the visible nodes in a Boltzmann machine are those to which I feed input data, and from which I receive the result of the network, that is, the generated data?

I found the following description of a fully connected Boltzmann machine: https://medium.com/@soumallya160/a-complete-guide-to-boltzmann-machine-deep-learning-7f7ce29b9e09 It says: "We have to fed into the hidden nodes". I'm not sure I understand this phrase correctly in the context of the guide.

Анатолий Глянцев
  • 11
  • 1

1 Answers1

0

Indeed visible nodes of RBMs are both where input data is provided and where the network's generative results can be observed. Hidden nodes, on the other hand, are not directly accessible and represent latent variables that model dependencies or capture higher-order patterns in the data.

Therefore the phrase "We have to fed into the hidden nodes" in the context of the article might be confusing and it really means that in this case the designer chose to represent known features like "Drama" and "Oscar-winning" as hidden nodes to simplify interpretation and directly associate movies with these features which helps to understand and explain recommendations in terms of familiar concepts, though it diverges slightly from the more abstract latent representations typically learned in RBMs.

cinch
  • 11,000
  • 3
  • 8
  • 17