everything should be done from scratch and in lisp
roadmap [3/15]
DONE implement decision tree learning
- State "DONE" from "TODO"
- State "TODO" from
DONE implement multivariate linear regression
- State "DONE" from "TODO"
- State "TODO" from
DONE implement feedforward neural networks
- State "DONE" from "TODO"
- State "TODO" from
GO implement a graphical interface for visual analysis
- State "GO" from "TODO"
- State "TODO" from
TODO abstraction over GPU APIs (CUDA, opencl)
- State "TODO" from
TODO implement convolutional neural networks (for image processing)
- State "TODO" from
TODO implement reinforcement learning (teach agents some physics?)
- State "TODO" from
maybe teach it to draw a circle play ping ping
TODO implement transformers
- State "TODO" from
TODO implement support vector machines
- State "TODO" from
TODO implement large language models (use llama and falcon-40b)
TODO build an AI companion
- State "TODO" from
TODO build an llm agent that modifies its own code
- State "TODO" from
TODO build real-world robotic agents (do i have the budget?)
- State "TODO" from
TODO study the philosophy of general intelligence
- State "TODO" from
TODO study latent spaces
- State "TODO" from
study guide
- decision tree
- multivariate linear regression
- feedforward neural network
- convolutional neural network