PSI 8: Deep Learning (extra)

In a GAN net, the interaction between the generator and the discriminator can be described as:

• A Laplace transform

• A zero sum game

• A Fourier transform

• A deepfake approximation

A Deep Q-Learning Net

• Uses multiple layers of Q-Learning to model a game player
• Applies Q-Learning to approximate the weights of a neural network
• Approximates a Q-Learning function using a neural network
• None of the others !

To enhance training a DQN we can use:

• Experience relay and Frozen the movie

• Faith and Luck

• Frozen the prediction network and experience relay

• Frozen the target network and experience replay

Some prompt methods to enhance the LLM results are:

• COP, LAMP, TOT, GET

• PoT, NoT, ToC, LoL

• CoT, L2MP, ToT, GoT

• CoT, LMP, T2T, G4T