Learning Transfer

Full Title or Meme

As humans moved from tribal societies to civilization one of the most important steps was the collection of learning of one generation and transferring that to a new generation.

Context

• In computer systems learning systems have similarly moved from always starting training with no knowledge base to transferring existing learning to new models as they are developed.

• The taking of skills and behavior learned in a training environment and transferring them to one's role is called Learning Transfer. Learning transfer is the goal of all training yet, it is surprisingly rare. Some estimates put the successful transfer rate of training between 10-22%. https://elearningindustry.com/need-talk-about-learning-transfer

Problems

• Moving from the first learning system where grandparents had the free time to transfer learning to children to a pedagogy where learning process themselves were developed into schools and eventually universities. That process took 10's of thousands of years.
• The pedagogy of today for humans is highly dependent of the entire learning system. France today, as one example, has a education system that was created by Napoleon to replace the wild variation in educational systems that existed prior to his consolship. This process started with mandating that all education be in the French spoken in Paris. No such common language exists for AI systems of today. Where a common language exists among AI, then an AI pedagogy can be created.

Solutions

The following are only the current state. Much more is to be done.

• Transfer Learning is a method in which a model developed for a particular task is used as a building block to solve a different problem. It is a domain of AI, which uses machine learning algorithms to improve learning capacities in one domain through previous exposure to another domain.