1. The Theory of Objects and the automatic generation of Intelligent Agents Sergio Pissanetzky 2009

2. What is an object, anyway? An object is anything that:  has information (attributes)  can use it to do something (behavior)  can share it with other objects (interaction).  organize information and reduce complexity.  fundamental structural element of information.  there can be no AI without a mastery of objects. What do they do? Our minds make objects all the time. We think, communicate, even dream in terms of objects.

3. Computer program.  behavior  structure  refactoring transformation  any system can be refactored But... where do they come from?

4.  Any finite system can be perfectly represented by a matrix of services in canonical form.  Under a deterministic, dissipative and chaotic dynamics, the structure of the system evolves towards its stable attractors.  The attractors are partitioned into submatrices that remain invariant under the dynamics.  The attractors represent the structure of the system.  The attractors also represent the equivalent logical circuit for the system.  The submatrices represent the objects present in the system, and their relationships represent the system’s ontology. The Matrix Theory of Objects establishes that:

5. Any system canonical matrix SCA ORA objects circuit The Matrix Model learning ontology

6. 3WGP 2 1AP 123 The Wumpus World

7. rulepercptruleCNFconclusionsdata ¬P 11 ¬B 11 B 11  P 12 V P 21 ¬B 11 V P 12 V P 21 ¬P 12 V B 11 ¬P 21 V B 11 ¬P 12 ¬P 21 S1CA S2CA S3CA S4AC S5AC S6AC S7AAC S8AAC 3WGP 2 1AP 123 ● Collect knowledge into the Learning Matrix. ● Organize the knowledge into canonical form. ● Minimize the profile to generate objects. CNF Resol

8. ¬B 11 ¬P 21 ¬P 12 V B 11 ¬P 11 B 11  P 12 V P 21 ¬P 12 ¬B 11 V P 12 V P 21 ¬P 21 V B 11 data S7AAC S5CA S1CA S8ACA S4AC S3CA S2CA S6AC 3WGP 2 1AP 123 Information obtained by the Learning Matrix may be incoherent, disorganized.

9. B 11  P 12 V P 21 ¬B 11 ¬P 11 ¬P 12 V B 11 ¬P 21 V B 11 ¬B 11 V P 12 V P 21 ¬P 12 ¬P 21 data S3CA S2CA S1CA S5AC S6AC S4AC S7AAC S8AAC 3WGP 2 1AP 123 Profile = 40 The Sorter organizes the information into a canonical form But there there are still no objects

10. ¬P 11 B 11  P 12 V P 21 ¬B 11 V P 12 V P 21 ¬P 12 V B 11 ¬P 21 V B 11 ¬B 11 ¬P 12 ¬P 21 data S1CA S3CA S4AC S5AC S6AC S2CA S7AAC S8AAC 3WGP 2 1AP 123 SCA has automatically refactored the matrix and has generated the following attractor Profile = 23 But objects are still difficult to discern

11. ¬P 11 B 11  P 12 V P 21 ¬B 11 V P 12 V P 21 ¬P 12 V B 11 ¬P 21 V B 11 ¬B 11 ¬P 12 ¬P 21 data S1CA S3CA S4AC S5AC S6AC S2CA S7AAC S8AAC 3WGP 2 1AP 123 X Y Based on a statistical sample of attractors, ORA has automatically generated two objects Profile = 23 The two objects are intelligent agents

12. inference rules biconditional elimination implication elimination De Morgandistributivityresolution The ontology of the system generated by SCA/ORA Y (reason) X (CNF)

13. data¬P 11 B 11  P 12 V P 21 ¬B 11 V P 12 V P 21 ¬P 12 V B 11 ¬P 21 V B 11 ¬B 11 ¬P 12 ¬P 21 S8S8 S7S7 S6S6 S5S5 S4S4 S3S3 S1S1 S2S2 The equivalent logical circuit of the system generated by SCA/ORA