Chapter
The Brain's Limitations on Neural Networks
36:17 - 44:57 (08:39)
The limitations of the brain's short range connectivity, paired with the limited number of long range connectivity, result in the construction of deep convolutional networks. Stochastic gradient descent is what artificial neural networks use to learn and solve the problems presented, however, for images and other physical compositions, more data is needed than the amount of parameters used.
Clips
The limited long range connectivity in biological neural networks necessitates the construction of deep convolutional neural networks.
36:17 - 44:10 (07:52)
Summary
The limited long range connectivity in biological neural networks necessitates the construction of deep convolutional neural networks. In artificial neural networks, stochastic gradient descent is used to adjust parameters to solve problems based on labeled data.
ChapterThe Brain's Limitations on Neural Networks
EpisodeTomaso Poggio: Brains, Minds, and Machines
PodcastLex Fridman Podcast
The over parameterization of solutions makes it easier to find good solutions, but it's possible that there are a lot of those solutions that aren't very good.
44:10 - 44:57 (00:46)
Summary
The over parameterization of solutions makes it easier to find good solutions, but it's possible that there are a lot of those solutions that aren't very good.