Clicking Video Tutorial returns 404

[U-Sahaj]

Please advise.
Thank you.

[emrekiciman]

Thanks for finding that. The video is reachable via https://www.youtube.com/watch?v=LALfQStONEc&t=114s I'll update the link

[U-Sahaj]

Thank you!

Have you come across this paper - https://doi.org/10.1038/s42256-018-0005-0?

How do you think it is related to DoWhy?

[emrekiciman]

I'm not familiar with that work. What connection do you see?

[U-Sahaj]

As its abstract says ...

Complex behaviour emerges from interactions between objects produced by different generating mechanisms. Yet to decode their causal origin(s) from observations remains one of the most fundamental challenges in science. Here we introduce a universal, unsupervised and parameter-free model-oriented approach, based on the seminal concept and the first principles of algorithmic probability, to decompose an observation into its most likely algorithmic generative models. Our approach uses a perturbation-based causal calculus to infer model representations. We demonstrate its ability to deconvolve interacting mechanisms regardless of whether the resultant objects are bit strings, space–time evolution diagrams, images or networks. Although this is mostly a conceptual contribution and an algorithmic framework, we also provide numerical evidence evaluating the ability of our methods to extract models from data produced by discrete dynamical systems such as cellular automata and complex networks. We think that these separating techniques can contribute to tackling the challenge of causation, thus complementing statistically oriented approaches.

the paper is introducing a universal, unsupervised and parameter-free model-oriented approach to decompose an observation into its most likely algorithmic generative models. meaning it is doing automatic causal factors discovery without any prior assumptions about the distribution of any variables involved, yes? The word deconvolve sounds really intriguing, isn't it? It kinds of do step #1 of the 4 steps in DoWhy, yes? And I believe it can complement DoWhy which I am not wrong is a statistically oriented approach beside the human expert input bit, yes?

Your thoughts?

[simplexityware]

Have I missed your reply somehow?