English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Data used to train machine learning models can be adversarial--maliciouslyconstructed by adversaries to fool the model. Challenge also arises by privacy,confidentiality, or due to legal constraints when data are geographicallygathered and stored across multiple learners, some of which may hold even an"anonymized" or unreliable dataset. In this context, the distributionallyrobust optimization framework is considered for training a parametric model,both in centralized and federated learning settings. The objective is to endowthe trained model with robustness against adversarially manipulated input data,or, distributional uncertainties, such as mismatches between training andtesting data distributions, or among datasets stored at different workers. Tothis aim, the data distribution is assumed unknown, and lies within aWasserstein ball centered around the empirical data distribution. This robustlearning task entails an infinite-dimensional optimization problem, which ischallenging. Leveraging a strong duality result, a surrogate is obtained, forwhich three stochastic primal-dual algorithms are developed: i) stochasticproximal gradient descent with an $\epsilon$-accurate oracle, which invokes anoracle to solve the convex sub-problems; ii) stochastic proximal gradientdescent-ascent, which approximates the solution of the convex sub-problems viaa single gradient ascent step; and, iii) a distributionally robust federatedlearning algorithm, which solves the sub-problems locally at different workerswhere data are stored. Compared to the empirical risk minimization andfederated learning methods, the proposed algorithms offer robustness withlittle computation overhead. Numerical tests using image datasets showcase themerits of the proposed algorithms under several existing adversarial attacksand distributional uncertainties.

Learning while Respecting Privacy and Robustness to Distributional Uncertainties and Adversarial Data