High Fidelity Image Counterfactuals with Probabilistic Causal Models — mentioned by slothofgalaxy

Wed, 26 Jul 2023 19:16:05 GMT

Ever wondered how we can make our deep generative models respect causal structure? This is key to creating authentic "what if" scenarios in our images! In our latest research, we deal with high-fidelity image counterfactuals, the generation of images based on "what if" scenarios that align with a specified causal graph. Why is this important? Causality gives us the tools to carry out principled counterfactual inference, which - among other things - is useful for maintaining subject identity in image counterfactuals. Principled counterfactuals of structured variables like images have great…

Paper and discussion history. A mention is not an endorsement.

High Fidelity Image Counterfactuals with Probabilistic Causal Models · HackerPapers discussions

High Fidelity Image Counterfactuals with Probabilistic Causal Models — mentioned by slothofgalaxy