Distributed Feature Selection for Multi-Class Classification Using ADMM

Feature selection is an important task in data-driven control applications to identify relevant features and remove non-informative ones, for example residual fault diagnosis. For multi-class data, the objective find a minimal set of that can distinguish data from all different classes. A distributed feature algorithm derived using convex optimization Alternating Direction Method Multipliers. The scales well with increasing number classes by utilizing parallel computations. Two case studies are used evaluate developed algorithm: classification internal combustion engine MNIST illustrate larger problem.