Sparse GPs with the collapsed bound (as in Titsias, 2009)

Questions
1

I wanted to learn inducing points by optimising the evidence lower bound in Titsias, (2009).

I guess, setting a flat prior on Xu and doing MAP is as close as one can get to that?

Xu_init = 10 * np.random.rand(20)

with pm.Model() as model:
    ℓ = pm.Gamma("ℓ", alpha=2, beta=1)
    η = pm.HalfCauchy("η", beta=5)

    cov = η ** 2 * pm.gp.cov.Matern52(1, ℓ)
    gp = pm.gp.MarginalSparse(cov_func=cov, approx="VFE")

    # set flat prior for Xu
    Xu = pm.Flat("Xu", shape=20, testval=Xu_init)

    σ = pm.HalfCauchy("σ", beta=5)
    y_ = gp.marginal_likelihood("y", X=X, Xu=Xu[:, None], y=y, noise=σ)

    mp = pm.find_MAP()
2 Likes