From Bilinear Regression to Inductive Matrix Completion: A Quasi-Bayesian Analysis

. 2023 Feb 11;25(2):333. doi: 10.3390/e25020333

Algorithm A2 MALA

Input: The data.
Parameters: Positive real numbers $λ, τ, h, T$
Output: The matrix $\hat{M}$
Initialize: $M_{0}; \hat{M} = 0_{m \times p}$
for $k = 1$ to T do
Sample ${\tilde{M}}_{k}$ from (9).
Set $M_{k} = {\tilde{M}}_{k}$ with probability $A_{M A L A}$ , from (10), otherwise $M_{k} = M_{k - 1}$ .
$\hat{M} \leftarrow \hat{M} + M_{k} / T$ .
end for