Errata: At 35:56 : I used a wrong sign. It is already highlighted in the video, the PDF on Github is fixed: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf At 39:06 : I also used a wrong sign. In the Schur complement before, it was already correct. I seemed to have made a mistake copying it. This is not highlighted in the video. Thanks @David Park for pointing this out. The pdf over on GitHub contains a remark and has been fixed accordingly: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf
You're very welcome :). Back when I created the video, I also always wanted a video in which someone walked me through all the nitty-gritty details. Thanks a lot for appreciating this!
Thank you for explaining this step by step! I also like that you provide nice visualisations along with the explanation, since I can understand things much better when I can see them
Truly appreciated all of your great contents! Just to clarify, in the summary session, you wrote "Sigma a|b = Sigma aa + Sigma ab * Sigma bb inverse * Sigma ba". Shouldn't it be "Sigma a|b = Sigma aa - Sigma ab * Sigma bb inverse * Sigma ba.", minus in between? In any case, I am truly impressed, and enjoy your thoughtful step-by-step detailed lectured! Thanks!
Thanks for the comment! :) Good catch! You are absolutely correct, it should be with a minus in the summary. It's a bit strange to me why I copied it that way, since I had it already correct above in the definition of the Schur complement. Nevertheless, I fixed the PDF over on Github: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf I will leave a pinned comment, thanks for pointing this out 😊. And also thanks a lot for the kind words.
You're very welcome :) Thanks for enjoying it. It's nice to hear, you appreciate the step-by-step approach. It is something that was missing on RUclips and it helped me a lot myself to go through the derivations in that detail.
Hey, I definitely have videos on Gaussian Processes planned for the future. Unfortunately, I can't give you a time estimate yet. It is a little lower on my priority atm. First, I want to continue with some Variational Inference, then Normalizing Flows and interleaved with other topics of the channel. Maybe towards the fall of this, the Playlist reaches the point to focus on Gaussian Processes. :)
That's of course a valid question. 😁 It is going to be important, once we look at Gaussian Process Regression and comes in handy at some other derivations. I will link the videos here, once they go online. (coming in the next months)
Errata:
At 35:56 : I used a wrong sign. It is already highlighted in the video, the PDF on Github is fixed: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf
At 39:06 : I also used a wrong sign. In the Schur complement before, it was already correct. I seemed to have made a mistake copying it. This is not highlighted in the video. Thanks @David Park for pointing this out. The pdf over on GitHub contains a remark and has been fixed accordingly: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf
Thank you so much for explaining every step, I'm sure there was a part of you that wanted to skip over the expansions.
You're very welcome :).
Back when I created the video, I also always wanted a video in which someone walked me through all the nitty-gritty details. Thanks a lot for appreciating this!
Thank you for explaining this step by step! I also like that you provide nice visualisations along with the explanation, since I can understand things much better when I can see them
You're very welcome! :)
It's the same for me with the visualizations.
Truly appreciated all of your great contents! Just to clarify, in the summary session, you wrote "Sigma a|b = Sigma aa + Sigma ab * Sigma bb inverse * Sigma ba". Shouldn't it be "Sigma a|b = Sigma aa - Sigma ab * Sigma bb inverse * Sigma ba.", minus in between? In any case, I am truly impressed, and enjoy your thoughtful step-by-step detailed lectured! Thanks!
Thanks for the comment! :)
Good catch! You are absolutely correct, it should be with a minus in the summary. It's a bit strange to me why I copied it that way, since I had it already correct above in the definition of the Schur complement. Nevertheless, I fixed the PDF over on Github: github.com/Ceyron/machine-learning-and-simulation/blob/main/english/essential_pmf_pdf/multivariate_normal_marginal_and_conditional.pdf
I will leave a pinned comment, thanks for pointing this out 😊. And also thanks a lot for the kind words.
This video was really helpful for me. Thank you for explaining this step by step!
You're very welcome :)
Thanks for enjoying it. It's nice to hear, you appreciate the step-by-step approach. It is something that was missing on RUclips and it helped me a lot myself to go through the derivations in that detail.
This video is so useful, it deserves much more views
Thanks a lot for the kind words 😊
Feel free to share it with your colleagues and peers. I would extremely appreciate that.
Thanks for the great explanation,
You're welcome. :)
Thanks for the kind words.
Thanks a lot for this video! You are awesome :)
You're very welcome!
Can we look at Gaussian Process for a future video given the connection with this video.
Hey,
I definitely have videos on Gaussian Processes planned for the future. Unfortunately, I can't give you a time estimate yet. It is a little lower on my priority atm. First, I want to continue with some Variational Inference, then Normalizing Flows and interleaved with other topics of the channel. Maybe towards the fall of this, the Playlist reaches the point to focus on Gaussian Processes. :)
Sir please add one practical question related this topic
Hi,
what do you mean by a practical question, like an application? You would find something like this in Gaussian Process Regression :).
@@MachineLearningSimulation yes,like application using numerical figures..
Just trying to see when something like this can be useful to apply.
That's of course a valid question. 😁
It is going to be important, once we look at Gaussian Process Regression and comes in handy at some other derivations. I will link the videos here, once they go online. (coming in the next months)