locked
Model derivation with Infer.NET RRS feed

  • Question

  • Hello,

    I was wondering if it is possible to use Infer.NET as an inference engine in order to derive new high-level concepts from existing concepts in my model.

    Suppose for example I have a model containing facts: Add(subClassOf, B, A), Remove(subclassOf, B, D)... After inference, I would be able to retrieve ChangeParent(B, A, D).

    My apologies if my question seems unclear. I am not familiar with the domain of inferencing, reasoning.

    Thanks.

    Friday, April 11, 2014 2:55 PM

All replies

  • Hi Bram,

    Infer.NET is a framework for running inference in probabilistic models. Changing the parent of B from A to D sounds like a deterministic operation to me. Or are these objects random variables drawn from some (discrete) distribution? If you can express your problem as a factor graph, then, yes, Infer.NET is extensible - you can add new factors, operators and distributions.

    Cheers,
    Yordan

    Saturday, April 12, 2014 12:12 PM