https://oteret.github.io/ Recent content on Ettore's blog Hugo -- gohugo.io en etr@google.com (Ettore Randazzo) etr@google.com (Ettore Randazzo) ©{year}, All Rights Reserved Fri, 21 Jul 2023 00:00:00 +0000 https://oteret.github.io/posts/biomaker-ca/ Fri, 21 Jul 2023 00:00:00 +0000 etr@google.com (Ettore Randazzo) Fri, 21 Jul 2023 00:00:00 +0000 https://oteret.github.io/posts/biomaker-ca/ Work done at Google, in collaboration with Alexander Mordvintsev. For an explanation of the project, see this web article or this paper. Ettore Randazzo featured image https://oteret.github.io/posts/isotropic-ca/ Tue, 11 Jul 2023 00:00:00 +0000 etr@google.com (Ettore Randazzo) Tue, 11 Jul 2023 00:00:00 +0000 https://oteret.github.io/posts/isotropic-ca/ Work done at Google, in collaboration with Alexander Mordvintsev, Craig Fouts, and Eyvind Niklasson. See the article. Ettore Randazzo featured image https://oteret.github.io/posts/particle-lenia/ Fri, 23 Dec 2022 00:00:00 +0000 etr@google.com (Ettore Randazzo) Fri, 23 Dec 2022 00:00:00 +0000 https://oteret.github.io/posts/particle-lenia/ Work done at Google, in collaboration with Alexander Mordvintsev and Eyvind Niklasson. See the article. Ettore Randazzo featured image https://oteret.github.io/posts/self-repl-nn/ Wed, 21 Jul 2021 00:00:00 +0000 etr@google.com (Ettore Randazzo) Wed, 21 Jul 2021 00:00:00 +0000 https://oteret.github.io/posts/self-repl-nn/ Work done at Google, in collaboration with Luca Versari and Alexander Mordvintsev. Can we create feedforward neural networks that can both encode a target image and self-replicating with and without variation? Yes… after quite some observations… Work published and presented at ALIFE 2021. See the paper. Ettore Randazzo featured image https://oteret.github.io/posts/adversarial-ca/ Fri, 07 May 2021 00:00:00 +0000 etr@google.com (Ettore Randazzo) Fri, 07 May 2021 00:00:00 +0000 https://oteret.github.io/posts/adversarial-ca/ Work done at Google, in collaboration with Alexander Mordvintsev, Eyvind Niklasson and Michael Levin. Hijacking the behaviour of pretrained Neural CA with adversarial attacks. See the Distill article. Ettore Randazzo featured image https://oteret.github.io/posts/mnist-ca/ Thu, 27 Aug 2020 00:00:00 +0000 etr@google.com (Ettore Randazzo) Thu, 27 Aug 2020 00:00:00 +0000 https://oteret.github.io/posts/mnist-ca/ Work done at Google, in collaboration with Alexander Mordvintsev, Eyvind Niklasson, Michael Levin and Sam Greydanus. End-to-end differentiable Neural CAs capable of self-classifying the MNIST digit they compose. See the Distill article and the colab for reproducing all the experiments. Ettore Randazzo featured image https://oteret.github.io/posts/mplp/ Fri, 03 Jul 2020 00:00:00 +0000 etr@google.com (Ettore Randazzo) Fri, 03 Jul 2020 00:00:00 +0000 https://oteret.github.io/posts/mplp/ Work done at Google, in collaboration with Eyvind Niklasson and Alexander Mordvintsev. Meta learning a learning protocol that can be used instead of SGD, Adam, etc, to train feed-forward neural networks. At least, in theory. See the Arxiv preprint and the code base. The code base contains colabs that show how to use it! My first first author work. Very fundamental in nature and small in scope. Still, it is 9 pages long. Ettore Randazzo featured image https://oteret.github.io/posts/growing-ca/ Tue, 11 Feb 2020 00:00:00 +0000 etr@google.com (Ettore Randazzo) Tue, 11 Feb 2020 00:00:00 +0000 https://oteret.github.io/posts/growing-ca/ Work done at Google, in collaboration with Alexander Mordvintsev, Eyvind Niklasson, and Michael Levin. See the Distill article. Ettore Randazzo featured image