How to to write multi-lined subscript in latex?

In latex, what is the best way to write multi-lined formula in subscript? Use `\atop`! For example, \begin{align*} P(M_n=1) =\sigma(&w_s(\sum\limits_{j_r \in T_{nr}}\vec{U_{j_r}} \cdot \vec{C_{j_r}} – \sum\limits_{j_b \in T_{nb}}\vec{U_{j_b}} \cdot \vec{C_{j_b}} ) + \\ &w_b(\sum\limits_{j_{r_1}, j_{r_2} \in T_{nr} \atop j_{r_1} \neq j_{r_2} }\left\Vert\vec{C_{j_{r_1}}} – \vec{C_{j_{r_2}}}\right\Vert^2 – \sum\limits_{j_{b_1}, j_{b_2} \in T_{nb} \atop j_{b_1} \neq j_{b_2} }\left\Vert\vec{C_{j_{b_1}}} – …

How to write multi-lined formula?

In latex, what is the best way to write multi-lined formula? Actually, the `begin/end{align*}` block supports `\\\` in it. For example, \begin{align*} P(M_n=1) =\sigma(&w_s(\sum\limits_{j_r \in T_{nr}}\vec{U_{j_r}} \cdot \vec{C_{j_r}} – \sum\limits_{j_b \in T_{nb}}\vec{U_{j_b}} \cdot \vec{C_{j_b}} ) + \\ &w_b(\sum\limits_{j_{r_1}, j_{r_2} \in T_{nr} \atop j_{r_1} \neq j_{r_2} }\left\Vert\vec{C_{j_{r_1}}} – \vec{C_{j_{r_2}}}\right\Vert^2 – \sum\limits_{j_{b_1}, j_{b_2} \in T_{nb} \atop j_{b_1} …

Minimum Description Length (MDL): a scoring function to learn Bayesian Network Structure

Bayesian Network Augmented Naive-Bayes (BAN) is an improved version of Naive-Bayes, in which every attribute $latex X_i$ may have at most one other attribute as its parent other than the class variable $latex C$. For example (figure from [1]): Though BAN provides a more flexible structure compared to Naive Bayes, the structure (i.e., dependence/independence relationships …