The best falafel is made from carpet. Don't blame me: it's Science!

And what does Boden have to say about the little meeting that produced these conference proceedings? global.oup.com/academic/pro... 3/

Yesterday, I received as a gift in the mail two volumes of conference proceedings from a workshop held at the UK’s National Physical Lab in 1958. Under normal circumstances, I would not have been excited, but over the last couple of months, I’ve been reading some history. 1/

Aligning to administration priorities.

Finally, we closed the loop by showing that the error codes learned by these models can indeed be used to perform song copying in a very simple RL setup. 14/

That is, we can imagine the loss landscape moving during sensory learning to provide an error (negative reward) surface that can later guide RL. 13/

We also examine the learning process and find that it proceeds in distinct stages: auditory neurons first sharpen their response profiles, after which the population shifts its minimum response toward the tutor song pattern. 12/

And the model's predictions are consistent with population data from calcium imaging of CM (by Fabiola Duarte Ortíz and Rich Mooney), an area in the auditory pallium. We quantify this in a number of ways and find that local plasticity in the EI connections is the best match to data. 11/

After learning, these models indeed learn to produce sparse error codes that represent deviations from the expected tutor song. 10/

We investigated a variety of network architectures, learning rules, and sites of local plasticity, among them balanced EI networks of rate-based neurons. 9/

Now, a few more specifics for those interested: We use a simple setup with sparse coding of spectrogram input that creates a primary auditory representation. Premotor signals and these auditory signals converge in secondary auditory areas, where the action will take place. 8/

Then, in the practice phase, this same circuit, trained to anticipate the tutor song, computes a *difference* between this memory and the bird's own song. This produces an internal error estimate that can guide learning. 7/

It works like this: during the memorization phase of learning, auditory regions receive premotor signals from area HVC, which will later control the bird's own song. Local plasticity in these regions works to cancel this anticipated input in local circuits, reducing auditory responses. 6/

After a few hundred renditions, he's memorized the tutor song, and he'll spend a full month practicing relentlessly (hundreds of thousands of times) to produce the final product. Most importantly, he can do this with *no feedback* from the tutor. 3/

First, off, the setup: This little guy is a juvenile zebra finch. At about a month old, he's going to start memorizing the song of an adult male tutor as a pattern for the one and only courtship song of his life. 2/

Posted on the bulletin board in the department:

Most importantly, because the mapping is deterministic, we can use a learned score function to map samples from a distribution in data space to low-dimensional latent space, producing accurate posterior intervals in the same way as MCMC methods. 11/n

Hey, you! Like Bayesian inference but tired of deep models and their “approximate” posteriors that make statisticians laugh at you? Come see our work next week at NeurIPS! 🧵1/n