Structured SVM - Separation

Separation

The above quadratic program involves a very large, possibly infinite number of linear inequality constraints. In general, the number of inequalities is too large to be optimized over explicitly. Instead the problem is solved by using delayed constraint generation where only a finite and small subset of the constraints is used. Optimizing over a subset of the constraints enlarges the feasible set and will yield a solution which provides a lower bound on the objective. To test whether the solution violates constraints of the complete set inequalities, a separation problem needs to be solved. As the inequalities decompose over the samples, for each sample the following problem needs to be solved.

y_n^* = \underset{y \in \mathcal{Y}}{\textrm{argmax}} \left( \Delta(y_n,y) + \boldsymbol{w}'\Psi(\boldsymbol{x}_n,y) - \boldsymbol{w}'\Psi(\boldsymbol{x}_n,y_n) - \xi_n\right)

The right hand side objective to be maximized is composed of the constant and a term dependent on the variables optimized over, namely . If the achieved right hand side objective is smaller or equal to zero, no violated constraint for this sample exist. If it is strictly larger than zero, the most violated constraint with respect to this sample has been identified. The problem is enlarged by this constraint and resolved. The process continues until no violated inequalities can be identified.

If the constants are dropped from the above problem, we obtain the following problem to be solved.

This problem looks very similar to the inference problem. The only difference is the addition of the term . Most often, it is chosen such that it has a natural decomposition in label space. In that case, the influence of can be encoded into the inference problem and solving for the most violating constraint is equivalent to solving the inference problem.

Read more about this topic:  Structured SVM

Famous quotes containing the word separation:

    “I was the one who was working to destroy the one thing to which I was committed, that is, my relationship with Gilberte; I was doing so by creating, little by little and through the prolonged separation from my friend, not her indifference, but my own. It was toward a long and cruel suicide of the self within myself which loved Gilberte that I continuously set myself ...”
    —Marcel Proust (1871–1922)

    “Just as children, step by step, must separate from their parents, we will have to separate from them. And we will probably suffer...from some degree of separation anxiety: because separation ends sweet symbiosis. Because separation reduces our power and control. Because separation makes us feel less needed, less important. And because separation exposes our children to danger.”
    —Judith Viorst (20th century)