One such technique seems to be non-default classification. For example, an AGI might need to learn to recognise new object categories as being newly-discovered object categories and the same object from different angles as being in the same object category. This technique seems important to be almost universally used for neural networks in an AGI brain. An AGI should be able to recognise new genres, new situations, new actions, new solutions, new consequences and new personality types in humans, just to name a few.
Are there any other techniques that would be used for multiple neural networks in an AGI brain?
Thank you for reading. 