The main objective of this project is to investigate, develop and evaluate state-of-the-art deep-learning algorithms for joint few-shot classification and out-of-distribution (OOD) detection. Few-shot learning deals with the challenges of limited supervision, and OOD detection attempts to identify inputs that do not belong to the set of classes seen during training. The two research problems are in line with several applications that are of high interest to the industrial partner as they tackle realistic open-set and limited-supervision scenarios. The specific technical objectives of this proposal are: (1) building a realistic few-shot learning benchmark, which reflects realistic open-set settings, with the possible emergence of completely unseen classes, out-of-distribution samples, domain shifts and imbalanced class distributions; (2) investigating and developing non-parametric mode-finding approaches for joint few-shot classification and OOD detection; and (3) investigating and developing domain-adaptation strategies and customized loss functions, which leverage unlabeled data from various domains during training, to mitigate the domain-shift challenges often encountered in industrial settings. The project will involve one intern (a postdoctoral fellow), whose objective is to advance the state-of-the-art in few-shot learning and OOD detection, while accounting for specific challenges and applications that are of interest to the industrial partner.