Adversarially-Trained Nonnegative Matrix Factorization
This work addresses robustness in latent dimensionality reduction for applications like recommendation systems, but it is incremental as it builds on existing adversarial training and NMF frameworks.
The paper tackles the problem of improving nonnegative matrix factorization's robustness and generalization by incorporating adversarial training, where an attacker adds bounded perturbations to the data, and demonstrates superior predictive performance in matrix completion tasks compared to state-of-the-art methods.
We consider an adversarially-trained version of the nonnegative matrix factorization, a popular latent dimensionality reduction technique. In our formulation, an attacker adds an arbitrary matrix of bounded norm to the given data matrix. We design efficient algorithms inspired by adversarial training to optimize for dictionary and coefficient matrices with enhanced generalization abilities. Extensive simulations on synthetic and benchmark datasets demonstrate the superior predictive performance on matrix completion tasks of our proposed method compared to state-of-the-art competitors, including other variants of adversarial nonnegative matrix factorization.