CSSE 413: Day 7

Outline

• Discussion of Deep Blue
• Engineered Intelligence (EI)
• A class experiment
• Bias nodes
• Basic properties of data sets:
  • Data set and NN are appropriately sized
  • Input of data set captures the domain to be modeled, avoids OOD cases
  • Desired output of data set has desired distribution.

Resources

• History of computer chess software
• IBM Research scientist Dr. Murray Campbell on Deep Blue
• Pattern recognition in NN-based AI
• Training, testing, and validation sets
• Learning AND with sigmoid and bias node
• Inputs and desired outputs of a skewed data set.
• Inputs and desired outputs of a more realistic data set
• LLMs that want to kill
• Fixing LLMs that want to kill by balancing data set