Self-classifying MNIST Digits (Paper Explained)

#ai #biology #machinelearning Neural Cellular Automata are models for how living creatures can use local message passing to reach global consensus without a central authority. This paper teaches pixels of an image to communicate with each other and figure out as a group which digit they represent. On the way, the authors have to deal with pesky side-effects that come from applying the Cross-Entropy Loss in combination with a Softmax layer, but ultimately achieve a self-sustaining, stable and continuous algorithm that models living systems. OUTLINE: 0:00 - Intro & Overview 3:10 - Neural Cellular Automata 7:30 - Global Agreement via Message-Passing 11:05 - Neural CAs as Recurrent Convolutions 14:30 - Training Continuously Alive Systems 17:30 - Problems with Cross-Entropy 26:10 - Out-of-Distribution Robustness 27:10 - Chimeric Digits 27:45 - Visualizing Latent State Dimensions 29:05 - Conclusion & Comments Paper: https://ift.tt/2EHCGFa My Video on Neural CAs: https://youtu.be/9Kec_7WFyp0 Abstract: Growing Neural Cellular Automata [1] demonstrated how simple cellular automata (CAs) can learn to self-organise into complex shapes while being resistant to perturbations. Such a computational model approximates a solution to an open question in biology, namely, how do cells cooperate to create a complex multicellular anatomy and work to regenerate it upon damage? The model parameterizing the cells’ rules is parameter-efficient, end-to-end differentiable, and illustrates a new approach to modeling the regulation of anatomical homeostasis. In this work, we use a version of this model to show how CAs can be applied to a common task in machine learning: classification. We pose the question: can CAs use local message passing to achieve global agreement on what digit they compose? Authors: Ettore Randazzo, Alexander Mordvintsev, Eyvind Niklasson, Michael Levin, Sam Greydanus Links: YouTube: https://www.youtube.com/c/yannickilcher Twitter: https://twitter.com/ykilcher Discord: https://ift.tt/3dJpBrR BitChute: https://ift.tt/38iX6OV Minds: https://ift.tt/37igBpB Parler: https://ift.tt/38tQU7C LinkedIn: https://ift.tt/2Zo6XRA If you want to support me, the best thing to do is to share out the content :) If you want to support me financially (completely optional and voluntary, but a lot of people have asked for this): SubscribeStar: https://ift.tt/2DuKOZ3 Patreon: https://ift.tt/390ewRH Bitcoin (BTC): bc1q49lsw3q325tr58ygf8sudx2dqfguclvngvy2cq Ethereum (ETH): 0x7ad3513E3B8f66799f507Aa7874b1B0eBC7F85e2 Litecoin (LTC): LQW2TRyKYetVC8WjFkhpPhtpbDM4Vw7r9m Monero (XMR): 4ACL8AGrEo5hAir8A9CeVrW8pEauWvnp1WnSDZxW7tziCDLhZAGsgzhRQABDnFy8yuM9fWJDviJPHKRjV4FWt19CJZN9D4n