Artificial intelligence based on cellular automata

Abstract

An artificial intelligence system can be implemented to identify relationships through the propagation of ripple patterns through a grid. In such a system, the grid may comprise cells which operate as cellular automata. Relationships may be identified based on collisions of signals detected by the cells in the grid, and, when a relationship is identified, it may be used to create high speed connections between cells.

Description

An artificial intelligence system can be implemented to identify relationships by propagating ripple patterns through a grid. In such a system, the grid can include cells as cellular automata. Relationships can be identified based on collisions of signals detected by cells in the grid, and once identified, they can be used to establish high-speed connections between cells. Abstract

Claims

1. A machine comprising a set of estimator modules, wherein each estimator module comprises: a) a random number generator; b) an observation input connection; c) circuitry configured to determine, based on an output of the random number generator, whether an observation signal on the observation input connection should be ignored for purposes of adjusting a number of observations associated with that estimator module; and d) circuitry configured to determine, based on the output of the random number generator, whether the observation signal on the observation input connection should be ignored for purposes of adjusting an observation value associated with that estimator module.

2. The machine of claim 1, wherein the machine comprises a set of estimators, wherein, for each estimator: a) that estimator comprises one or more estimator modules from the set of estimator modules; b) each estimator module comprised by that estimator is part of a chain which connects each estimator module comprised by that estimator and which does not include any estimator modules comprised by any other estimator; and c) for each estimator module comprised by that estimator: i) the number of observations associated with that estimator module is a number of observations corresponding to that estimator; and ii) the observation value associated with that estimator module is an observation value corresponding to that estimator.

3. The machine of claim 2, wherein the machine comprises circuitry configured to determine, for each estimator, the number of observations corresponding to that estimator based on how many estimator modules are comprised by that estimator.

4. The machine of claim 3, wherein, for each estimator from the set of estimators: a) each estimator module comprised by that estimator comprises: i) an extends in connection; ii) an extends out connection;iii) an extends out storage element, wherein the extends out storage element stores a value indicating if that estimator module is connected to another estimator module via the extends out connection; b) for each estimator module comprised by that estimator except for a head estimator module, that estimator module is connected to another estimator module comprised by that estimator via the extends in connection; and c) for each estimator module comprised by that estimator except for a tail estimator module, that estimator module is connected to another estimator module comprised by that estimator via the extends out connection.

5. The machine of claim 4, wherein, for each estimator from the set of estimators, for each estimator module comprised by that estimator, that estimator module comprises circuitry for determining, based on the output of the random number generator, whether the observation signal on the observation input connection should be ignored for purposes of adjusting the number of observations associated with that estimator module by determining whether to prevent a connection from being established from the tail estimator module comprised by that estimator to another estimator module via the extends out connection of the tail estimator module comprised by that estimator.

6. The machine of claim 5, wherein, for each estimator from the set of estimators, for each estimator module comprised by that estimator: a) that estimator module is configured to determine that it is the tail estimator module based on the extends out storage element storing a value indicating that estimator module is not connected to another estimator module via the extends our connection; b) that estimator module is configured to, based on: i) determining that the observation signal on the observation input connection should not be ignored for purposes of adjusting the number of observations associated with that estimator module; and ii) determining that that estimator module is the tail estimator module; connect to another estimator module via the extends out connection.

7. The machine of claim 4, wherein, for each estimator, for each estimator module comprised by that estimator:a) unless that estimator module is the tail estimator module, that estimator module is configured to: i) based on determining that the observation signal on the observation input connection should be ignored for purposes of adjusting the number of observations associated with that estimator module, send a no extend signal to the estimator module connected via the extends out connection, wherein the no extend signal indicates that the observation signal on the observation input connection should be ignored for purposes of adjusting the number of observations associated with that estimator module; and ii) based on receiving the no extend signal, send the no extend signal to the estimator module connected via the extends out connection; and b) that estimator module is configured to ignore the observation signal on the observation input connection for purposes of adjusting the number of observations associated with that estimator module based on receiving the no extend signal.

8. The machine of claim 3, wherein the circuitry configured to determine, for each estimator from the set of estimators, the number of observations corresponding to that estimator based on the number of estimator modules comprised by that estimator, is configured to, for each estimator from the set of estimators, calculate the number of observations corresponding to that estimator as an estimate equal to 2(m+1)^ 2, wherein m is a cardinality value of a set consisting of the one or more estimatorcomprised by that estimator.

9. The machine of claim 2, wherein: a) each estimator module from the set of estimator modules comprises an observation memory element; and b) the machine comprises circuitry configured to determine, for each estimator, the observation value corresponding to that estimator based on values stored in the observation memory elements comprised by that estimator.

10. The machine of claim 9, wherein a) for each estimator module from the set of estimator modules, the observation memory element comprised by that estimator module stores a binary value having a length equal to a uniform length; b) the circuitry configured to determine, for each estimator, the observation value corresponding to that estimator, is configured to make that determination by: i) defining a highest order portion of a binary representation of the observation value corresponding to that estimator as equal to a value stored in the observation memory element of a head estimator module comprised by that estimator; and ii) for each estimator module comprised by that estimator other than the head estimator module, defining a portion associated with that estimator module as equal to the value stored in the observation memory element of that estimator module, wherein the portion associated with that estimator module is a portion of the binary representation of the observation value corresponding to that estimator which is separated from the highest order portion by a product of the uniform length and a number of estimator modules separating that estimator module from the head estimator module in the chain which connects each estimator module comprised by that estimator.

11. The machine of claim 10, wherein the uniform length is one bit.

12. The machine of claim 10, wherein for each estimator, for each estimator module comprised by that estimator: a) that estimator module is configured to update the value stored in the observation memory element by making that value equal to observation signal on the observation input connection when that estimator module is first added to that estimator by being connected to another estimator module which is part of the chain connecting each estimator module comprised by that estimator; b) that estimator module comprises circuitry configured to update the value stored in the observation memory element by performing an inversion operation based on: i) receiving a signal from another estimator module comprised by that estimator indicating that no other estimator module had determined that theobservation signal on the observation input connection should be ignored for purposes of adjusting the observation value associated with that estimator module; ii) that estimator module determining that the observation signal on the observation input connection should not be ignored for purposes of adjusting the observation value associated with that estimator module; and iii) that estimator module determining that it is a tail estimator module in the chain connecting each estimator module comprised by that estimator; wherein the inversion operation changes the value stored in the observation memory element from an original value to an updated value by inverting the original value; c) that estimator module comprises circuitry configured to update the value stored in the observation memory element by performing the inversion operation based on receiving a carry signal from another estimator module comprised by the chain which connects each estimator module comprised by that estimator; and d) that estimator module comprises circuitry configured to send the carry signal to a higher order estimator module based on: i) completing the inversion operation; and ii) the original value of the value stored in the observation memory element not being equal to the observation signal on the observation input connection, wherein the higher order estimator module is comprised by the chain which connects each estimator module comprised by that estimator, and is separated from the tail estimator module by more estimator modules than that estimator module.

13. The machine of claim 11, wherein, for at least on estimator from the set of estimators: a) that estimator has an associated tree structure which comprises a set of nodes, wherein: i) the set of nodes comprises a subset of nodes, ii) for each node from the subset of nodes, that node is connected to a terminal set of estimator modules, wherein each estimator module from the terminal set of estimator modules connected to that node is comprised by that estimator; b) for each node from the set of nodes:i) except for a root node, each node from the set of nodes is connected to one parent node in the set of nodes; and ii) that node is connected to a set of children, wherein: A) for each node from the set of nodes, the set of children for that node comprises a first subset of children and a second subset of children, wherein the first subset of children and the second subset of children are disjoint; B) for each node from the subset of nodes, the set of children for that node comprises the set of estimator modules connected to that node; and C) for each node from the set of nodes other than nodes in the subset of nodes, the set of children for that node consists of other nodes from the set of nodes; and c) the machine comprises circuitry configured to, for any node on a shortest path between the root node and a target estimator module, based on one or more values stored by that node, identify a child of that node which is also on the shortest path between the root node and the target estimator module, for each of the following target estimator modules: i) an estimator module which stores a value representing a lowest order portion of the binary representation of the observation value corresponding to that estimator; ii) an estimator module which stores a value representing a lowest order 0 in the binary representation of the observation value corresponding to that estimator; and iii) an estimator module which stores a value representing a lowest order 1 in the binary representation of the observation value corresponding to that estimator.

14. The machine of claim 9, wherein: a) for each estimator module from the set of estimator modules, the observation memory element comprised by that estimator module is a tri-state memory element having a first state, a second state, and a third state; andb) the circuitry configured to determine, for each estimator, the observation value corresponding to that estimator, is configured to, for each estimator module which is part of the chain that connects each estimator module comprised by that estimator: i) treat the value stored in the observation memory element comprised by that estimator module as part of the observation value for that estimator when the observation memory element comprised by that estimator module is in the first state or the second state; ii) treat that estimator module as not part of that estimator when the observation memory element comprised by that estimator module is in the third state.

15. A machine comprising a set of segment cells, wherein each segment cell from the set of segment cells comprises: a) a set of power connections, wherein the set of power connections comprises first and second power connections, wherein each of the first and second power connections is configured to act as a power source for that segment cell; b) a set of signal connections; and c) circuitry configured to provide a set of output signals over the set of signal connections based on: i) a set of input signals received over the set of signal connections; and ii) which one or more power connections from the set of power connections are receiving power from outside of that segment cell.

Images