I got the chance to work on a new book with de Gruyter. More details follow when contributors are on board.
For mapping symbolic and subsymbolic data in neural nets in the symbolic world the available information must be structured available. Introducing a notation analog to Labannotation we can cover movements in wet-laboratories.
ARTICLE STILL UNDER CONSTRUCTION
Abstract. This paper discusses possible patterns of creating of levels of self-awareness in machines by implementing mechanisms like dreaming and self-observation, not by using knowledge.
It is still is asked what analogy actually dreams and consciousness have in the construction of thinking machines.
Selfconstruction of higher dimensions in ML. As the standard mechanism of learning is based on knowledge here might be the key to another construction pattern.
Move 37. In the construction of Alpha go there Google did a remarkable approach. They set a initial set of knowledge in the system (Recorded GO matches). But after that the cloned the AI and a few hundred thoughts began to play with each other. With this they created some kind of new BIAS different to human.
Claustrum. „There are three main types of cells.
The first, which is deemed Type 1, is large and covered with dendritic spines. These cells receive input as well as project back toward various regions, both laterally and medially.
The other two types of cells do not have spines, but can be told apart based on the cell body size. However, both are restricted to the claustrum and, thus, are labeled interneurons.“ (Wikipedia)