Sign in
Microsoft.com
 
Microsoft
 
 
 
Microsoft >
> 
> 
> 
Getting Started with Infer.Net

Unanswered Getting Started with Infer.Net

  • Friday, December 02, 2011 12:51 PM
     
     
    Sign In to Vote
    0

    Hi all,

    Just download infer.net and tested it by using the coin example from the pdc2009 video. It tested for 2 heads and I got the right answer. I would like to learn the use of infer.net but it appears you need some good probability backing so as to do something useful with it. I will like to learn with 2 examples. If someone can help me with the code for those 2, it will serve as a good place for me to start learning.

     

    1. Given a bag of 3 coins, 1 loaded (0.9H, 0.1T) and 2 normal(0.5H|T), if I pick a coin at random (1/3 chance of picking each coin), toss it and observe HHHTH, what is the probability that I picked the loaded coin.

    2. Given a bag  of 3 dice, 1 loaded (0.5(One), 0.1(Two-Six)) and 2 normal (1/6 of each value), if I pick a coin at random (1/3 chance of picking each coin), roll it and observer 1-1-1-5-1, what is the probability that I picked the loaded die?

    PS: The 2nd question is the same as the 1st except with dice so I can see how the code changes if it is not just true/false.

     

    Thanks in advance.

     

All Replies

  • Friday, December 02, 2011 8:22 PM
    Owner
     
     
    Sign In to Vote
    0

    Here is one way to implement this for 1.

    2 is similar but uses Discrete distributions and integer random variables.

    John

    // The model

    var numCoins = Variable.New<int>();

    var numTosses = Variable.New<int>();

    Range c = new Range(numCoins);

    Range t = new Range(numTosses);

    var coin = Variable.DiscreteUniform(c);

    var toss = Variable.Array<bool>(t);

    var coinPrior = Variable.Array<Bernoulli>(c);

    using (Variable.Switch(coin))

    {

      toss[t] = Variable<bool>.Random(coinPrior[coin]).ForEach(t);

    }

     

    // Observations. true = heads, baised coin is first one

    bool[] observedTosses = new bool[] { true, true, true, false, true };

    Bernoulli[] observedCoinPriors = new Bernoulli[] {

      new Bernoulli(0.9), new Bernoulli(0.5), new Bernoulli(0.5) };

     

    // Hook up the observations

    numCoins.ObservedValue = observedCoinPriors.Length;

    coinPrior.ObservedValue = observedCoinPriors;

    numTosses.ObservedValue = observedTosses.Length;

    toss.ObservedValue = observedTosses;

     

    // Run the inference

    InferenceEngine engine = new InferenceEngine();

    Discrete postCoin = engine.Infer<Discrete>(coin);

    Console.WriteLine("Probability I picked biased coin = {0:0.000}", postCoin.GetProbs()[0]);

     

     
  • Saturday, December 03, 2011 1:01 PM
     
     
    Sign In to Vote
    0

    Hi John,

    Thanks for the response. I'll try it and get back  to you.

     

    regards,