Multi-level hierarchical classification addresses the problem of classifying items into a multi-level hierarchy structure of classes. For example, an image of a ‘cat’’ can be classified into `biological organism’’, `animal’, and `cat’’, depending on the taxonomy used. While there have been several methods proposed to solve this problem, they still suffer from several drawbacks. In particular, most existing methods: (i) do not embed the taxonomy structure used, (ii) use a complex backbone neural network with ‘n$ disjoint output layers that do not constraint each other, and (iii) consequently, may output predictions that are often inconsistent with the taxonomy in place. This lecture addresses these deficiencies by introducing a novel mask-based output layer for multi-level hierarchical classification. Specifically, this lecture will cover a model-agnostic output layer that embeds the taxonomy and that can be combined with any model.