Joint Probability Trees
This provides a scalable and interpretable alternative to probabilistic graphical models for applications involving high-dimensional heterogeneous data, though it appears incremental as it builds on tree-based structures rather than introducing a new paradigm.
The authors tackled the challenge of making joint probability distribution learning and reasoning tractable for practical use by introducing Joint Probability Trees (JPTs), which handle both symbolic and subsymbolic variables without prior dependency assumptions, achieving linear scaling in learning and reasoning while supporting interpretable explanations for inferences.
We introduce Joint Probability Trees (JPT), a novel approach that makes learning of and reasoning about joint probability distributions tractable for practical applications. JPTs support both symbolic and subsymbolic variables in a single hybrid model, and they do not rely on prior knowledge about variable dependencies or families of distributions. JPT representations build on tree structures that partition the problem space into relevant subregions that are elicited from the training data instead of postulating a rigid dependency model prior to learning. Learning and reasoning scale linearly in JPTs, and the tree structure allows white-box reasoning about any posterior probability $P(Q|E)$, such that interpretable explanations can be provided for any inference result. Our experiments showcase the practical applicability of JPTs in high-dimensional heterogeneous probability spaces with millions of training samples, making it a promising alternative to classic probabilistic graphical models.