Computational Neuroscience/ML: Understanding Hopfield Networks

The Physics Nobel Prize 2024 was partially awarded to John Hopfield for his model of associative memory. In the next meetup event, let's come together and have a discussion on Hopfield Networks.

Why are Hopfield Networks interesting?

1. They are a model for memory that uses local learning rules. That's right: no backprop.
2. They use the biologically plausible Hebbian learning rule: "Neurons that fire together, wire together".
3. Certain continuous variants of Hopfield Networks are equivalent to self-attention.

You can learn more by reading this explainer blog post from Sepp Hochreiter's group. If you find that interesting, please join us for the evening.

Here's the plan for the evening

🍕 18:30: Pizza, drinks and networking
👋 19:00 - 19:30 Introductions
👨‍🏫 19:30 - 20:30: We cover classical Hopfield Networks in formal detail
💬 20:30 - 21:30: Casual round table discussion on Hopfield Networks and its relation to AGI

Hope to see you there!