Its quite nice to make this whole graph connect altogether in one go. As corny as it is.
At the core of my curiosity is the human brain. To me, it holds wonders more complex and interesting that the universe. That is why almost everything I’ve done so far has been an engineer’s deep dive into computationalNeuroscience .
The Biological
00 - Physiological Psychology Table of Contents 00 - General Psychology Table of Contents
… and a lot more silos that are everywhere in this dump
The Engineering
00 - Roboting 00 - C++ Table of Contents 00 - Deep Learning Table of Contents 00 - Multivariate Calculus Table of Contents 00 - Digital Control Systems Table of Contents
Lie Algebra What is Abstract Algebra (Modern Algebra) Topology Fourier Transform Singular Value Decomposition
… and a lot more silos that are everywhere in this dump
The Intersection
00 - Robot Embodiment 00 - Computational Neuroscience TOC
worldModeling action perception configuration endToEnd
Premise of Computational Neuroscience
We try to study the brain by making connections between it and modern algorithms. The goal is to make better algorithms as well as give us a better understanding of the brain.
The mouse brain is used more most of this course. Reason is that its the most completely studied animal brain, and the brain is relatively less complex than that of the brains of humans.
We can draw rough analogies to current algorithms and parts of the brain.
- Parts of the cortex :
- primarySomatosensoryCortex resembles most self-supervised learning selfSupervisedLearning
- primaryMotorCortex resembles optimal control (Control Systems)
- prefrontalCortex value functions for self-supervised learning
- Parts of the basalGanglia resemble concepts from reinforcementLearning
- olfactoryBulb scent feature engineering
- Parts of the hippocampus resembles memory formation (kinda part of worldModeling)
- cerebellum is a core part of stateEstimation and action
- hindbrain or the brainstem responsible for low-level functions for action
Why is machineLearning such a popular model of the brain?
Much of how we study the brain is our a property X contributes to a function Y. The way we study brain has kinda guided the development and modeling with ML.
ML in general is pretty good at approximating behavioral functions.