The rising interest around probabilistic modeling in Machine Learning and AI comes from the need to perform complex reasoning under uncertainty. For example, an agent operating in the real-world needs to answer complex probabilistic inference queries in order to make decisions and take actions. Typically, it is necessary to compute these answers in a limited amount of time. In many domains, such as healthcare and economic decision making, it is crucial that the result of these queries is reliable, i.e. either exact or coming with approximation guarantees. Tractable probabilistic models (TPMs) satisfy both of these two desiderata, as they allow for exact and efficient inference for a wide range of inference scenarios.

Recent years have shown, against wide-held belief, that TPMs are surprisingly powerful, and that one can indeed have both flexible modeling and exact poly-time inference. TPMs have been successfully used in image classification, completion and generation, activity recognition, language and speech modeling, bioinformatics, verification and diagnosis of physical systems, to name but a few. Examples of TPMs comprise, but are not limited to:
1. neural autoregressive models
2. normalizing flows
3. bounded-treewidth PGMs (probabilistic graphical models)
4. determinantal point processes
5. PGMs with high girth or weak potentials
6. exchangeable probabilistic models exploiting symmetries or some invariances
7. hinge-loss Markov random fields and probabilistic soft logics
8. probabilistic circuits (arithmetic circuits, sum-product networks, probabilistic sentential decision diagrams, cutset networks, etc.)

The TPM Workshop brings together researchers working on different fronts of this spectrum and aims to open the community to related disciplines --- from probabilistic programming to automated reasoning and verification. The focus of the 4th TPM workshop is to foster exchange within the community and the general UAI audience, by disseminating recent results and trends delivered by a diverse set of high-profile invited speakers.

Schedule: https://sites.google.com/view/tpm2021/schedule