Online Coregularization for Multiview Semisupervised Learning

Algorithm 1

A template online co-regularization algorithm for multiview semi-supervised binary classification problems. This template algorithm aims for increasing the dual function on each learning round.

INPUT: four positive scalars: 𝛾₁, 𝛾₂, 𝛾_𝑐 and 𝜇.

INITIALIZE: a coefficient vectors group 𝜶₁₀,𝜶₂₀,…,𝜶_𝑇0 and its associated

     decision boundary vectors 𝝎₀⁽¹⁾, 𝝎₀⁽²⁾.

PROCESS: For 𝑡=1,2,…,𝑇

      Receive an example (𝐱_𝑡^(1,2),𝑦_𝑡,𝜎_𝑡),

      Choose a new coefficient vectors group 𝜶_1𝑡,𝜶_2𝑡,…,𝜶_𝑇𝑡 that satisfies

         𝐷𝜶_1𝑡,𝜶_2𝑡,…,𝜶_𝑇𝑡≥𝐷𝜶_1𝑡−1,𝜶_2𝑡−1,…,𝜶_𝑇𝑡−1,

      Return two new associated decision boundary vectors 𝝎_𝑡⁽¹⁾, 𝝎_𝑡⁽²⁾

      using (19).