Think Deep

An approach to get computers to tink in a human like manner

Similar projects worth following
Today the standard process of getting robots to do something is to think about the task, phrase the solution in some programming language and put that on the robot's computer. The goal of this project is to give the computer information about itself and its environment and have it come up with the solution itself.

Consider this statement about thinking at 59:44 :

It is obvious that to think about something you need context. Context is basically the description of relations between things / description of things using relations (see graphical example in the gallery)

The goals of Think Deep are:

first stage:

  • Come up with a language to describe context easily
  • Create a program that can make sense of context and solve thinking exercises based on it

second stage:

  • implement ways to enable dynamic expansion of context - aka learning

  • Describing Context Using Text

    schuhumi08/23/2015 at 13:07 0 comments

    To get a computer to think it needs context. For now the primary way to put context into the computer will be a simple descriptive language.

    The bottom line I'll follow is: anything is something to something else. Example: "schuhumi" is "name" to "self", aka schuhumi is my name. "21" is "age" to "self", while "21" is "instance" to "number" as well. "21" is "increment" to "20". "age" is "property" of "self". "Birthday" is "manipulator" to "age" etc.

    My language to describe such things is very simple. First you declare all "something"s you have using a leading exclamation mark:

    ! self
    ! name
    ! schuhumi
    ! age
    ! number
    ! instance

    etc. Then you describe realtions using arrows:

    schuhumi -(name)-> self
    21 -(age)-> self
    21 -(instance)-> number
    20 -(increment)-> 21
    and so forth.

    As you can see you need A LOT of context to resemble human thinking - in fact the amout of context you get as a child while being raised and exploring the world yourself.

    Therefore experiments will start of with simulations of very simple robots, for example thinking about how to move to reach a certain destination (two dimensional).

View project log

Enjoy this project?



Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates