Our paper titled “Probabilistic Circuits That Know What They Don’t Know” has been accepted for publication at the Uncertainty in Artificial Intelligence (UAI) conference 2023 and has been selected for a 25 minute oral presentation.
The paper draws inspiration from the overconfidence phenomenon in neural networks, where false predictions on unknown concepts generally appear with a very large associated confidence in the (false) output. The paper first shows that this problem persists in probabilistic circuits, despite the models’ tractability. It then posits that this is due to a lack of uncertainty assessment in its probabilistic formulation. As a remedy, a closed-form solution to the popular Monte-Carlo dropout is derived, which provides an efficient assessment of uncertainty in the parameters in a computationally efficient way.
For more information, read the full paper or see the abstract below:
Probabilistic circuits (PCs) are models that allow exact and tractable probabilistic inference. In contrast to neural networks, they are often assumed to be well-calibrated and robust to out-of-distribution (OOD) data. In this paper, we show that PCs are in fact not robust to OOD data, i.e., they don’t know what they don’t know. We then show how this challenge can be overcome by model uncertainty quantification. To this end, we propose tractable dropout inference (TDI), an inference procedure to estimate uncertainty by deriving an analytical solution to Monte Carlo dropout (MCD) through variance propagation. Unlike MCD in neural networks, which comes at the cost of multiple network evaluations, TDI provides tractable sampling-free uncertainty estimates in a single forward pass. TDI improves the robustness of PCs to distribution shift and OOD data, demonstrated through a series of experiments evaluating the classification confidence and uncertainty estimates on real-world data.
Abstract of our paper “Probabilistic Circuits That Know What They Don’t Know”