Fig 11. Prior.

Intensity function λ(κ ∣ x, ξ) is expressed as the sum of kernels, and it is indexed by the discrete measure ξ defined on Θ^κ × Θ^x. This discrete measure ξ consists of atoms $\left\{({\mu }_k^{\kappa },{\Lambda }_k^{\kappa },{\theta }_k^x)\right\}$ and each ${\mu }_k^{\kappa },{\Lambda }_k^{\kappa },{\theta }_k^x$ corresponds to the mean fluorescent image, the precision matrix of the fluorescent images, and the tuning curve parameter of each neuron. A weight attached to each atom expresses the average firing rate of this neuron. We set the mixture of gamma processes ${\mathcal{G}}_K(\mathrm{d}\xi \mid {\alpha }_0,{\beta }_0,H,{\mathbb{u}}^x)$ as a prior for this ξ.