📗 -> Lecture Date: Name


Learning

🎤 Vocab

❗ Unit and Larger Context

Small summary

✒️ -> Scratch Notes

The “impossible task” of the error driven learning was solved with a hidden layer

  • Can be helpful to have smaller numbers of hidden neurons
    • Experiments with starting with small number of hidden neurons and increasing them has been experimented with

Overview of Learning

Biological:

  • Synaptic plasticity
    Computational
  • Self organizing - soaking up regularities
  • Error driven - focused on getting the right answers
    Math
  • Captures “Neurons that fire together, wire together”
  • (plus thresholds, phases)

🧪-> Example

Talking to that relative/person on the bus (someone outside the field) again: How can you use computer simulations to explore how we learn and become our unique selves, shaped by our collections of experiences?

Computer simulations can help to hone in on important functional components of our brain. For example, by trying to create a minimal representation of how a neuron fires, we can find what aspects we need / don't need in order to convey information well in machine learning, as well as what is important to understanding ourselves. We've found that we learn by association and through error, and the combination of these can be helpful in informing how we learn.

Resources

  • Put useful links here

Connections

  • Link all related words