US2019065936A1PendingUtilityA1

Anti-hebbian and hebbian computing with thermodynamic ram

53
Assignee: NUGENT ALEXPriority: Apr 4, 2014Filed: Jun 16, 2018Published: Feb 28, 2019
Est. expiryApr 4, 2034(~7.7 yrs left)· nominal 20-yr term from priority
Inventors:Alex Nugent
G06N 3/049G06N 3/063G06N 20/00G06N 99/005
53
PatentIndex Score
0
Cited by
0
References
0
Claims

Abstract

A thermodynamic RAM circuit composed of a group of AHaH (Anti-Hebbian and Hebbian) computing circuits that form one or more kT-RAM circuits. The AHaH computing circuits can be configured as an AHaH computing stack. The kT-RAM circuit(s) can include one or core kT-Cores, each partitioned into AHaH nodes of any size via time multiplexing. The kT-Core couples readout electrodes together to form a larger combined kT-Core. AHaH Computing is the theoretical space encompassing the capabilities of AHaH nodes. At this level of development solutions have been found for problems as diverse as classification, prediction, anomaly detection, clustering, feature learning, actuation, combinatorial optimization and universal logic.

Claims

exact text as granted — not AI-modified
1 . (canceled) 
     
     
         2 . (canceled) 
     
     
         3 . (canceled) 
     
     
         4 . (canceled) 
     
     
         5 . (canceled) 
     
     
         6 . (canceled) 
     
     
         7 . (canceled) 
     
     
         8 . A thermodynamic RAM (Random Access Memory) circuit, comprising:
 a collection of kT (Thermodynamic)-Core circuits, wherein each kT-Core circuit among said collection of core kT-Core circuits comprises an AHaH (Anti-Hebbian and Hebbian) circuit comprising an AHaH node circuit that provides a universal computational and adaptation resource, wherein at least one kT-Core circuit among said collection of kT-Core circuits couples readout electrodes together to form a larger combined kT-Core among said collection of kT-Core circuits.   
     
     
         9 . The circuit of  claim 8  further comprising an instruction set for a kT-Core learning circuit among said collection of kT-Core circuits. 
     
     
         10 . The circuit of  claim 8  further comprising at least one kT-RAM circuit that comprises at least one kT-Core circuit among said collection of said kT-Core circuits partitioned into AHaH nodes via temporal multiplexing. 
     
     
         11 . The circuit of  claim 9  further comprising at least one kT-RAM circuit that comprises at least one kT-Core circuit among said collection of said kT-Core circuits partitioned into AHaH nodes via temporal multiplexing. 
     
     
         12 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for classification. 
     
     
         13 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for prediction. 
     
     
         14 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for anomaly detection. 
     
     
         15 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for clustering. 
     
     
         16 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for feature learning. 
     
     
         17 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for actuation. 
     
     
         18 . The circuit of  claim 8  wherein said AHaH circuit is based on an AHaH node connection topology that provides for combinatorial optimization and universal logic. 
     
     
         19 . The circuit of  claim 8  wherein said AHaH circuit comprises at least one memristor. 
     
     
         20 . A thermodynamic RAM (Random Access Memory) circuit, comprising:
 a collection of kT (Thermodynamic)-Core circuits, wherein each kT-Core circuit among said collection of core kT-Core circuits comprises an AHaH (Anti-Hebbian and Hebbian) circuit comprising at least one memristor, wherein at least one kT-Core circuit among said collection of kT-Core circuits couples readout electrodes together to form a larger combined kT-Core among said collection of kT-Core circuits, wherein said AHaH circuit comprises an AHaH node circuit, and wherein said AHaH node circuit provides a universal computational and adaptation resource.   
     
     
         21 . The circuit of  claim 12  wherein said AHaH circuit is based on an AHaH node connection topology that provides for at least one of the following: classification, prediction, anomaly detection, clustering, feature learning, actuation, combinatorial optimization and universal logic. 
     
     
         22 . The circuit of  claim 12  further comprising an instruction set for a kT-Core learning circuit among said collection of kT-Core circuits. 
     
     
         23 . The circuit of  claim 12  further comprising at least one kT-RAM circuit that comprises at least one kT-Core circuit among said collection of said kT-Core circuits partitioned into AHaH nodes via temporal multiplexing. 
     
     
         24 . The circuit of  claim 12  wherein said AHaH circuit comprises an AHaH node circuit. 
     
     
         25 . The circuit of  claim 16  wherein said AHaH node circuit provides a universal computational and adaptation resource. 
     
     
         26 . A thermodynamic RAM (Random Access Memory) circuit, comprising:
 a collection of kT (Thermodynamic)-Core circuits, wherein each kT-Core circuit among said collection of core kT-Core circuits comprises an AHaH (Anti-Hebbian and Hebbian) circuit, wherein at least one kT-Core circuit among said collection of kT-Core circuits couples readout electrodes together to form a larger combined kT-Core among said collection of kT-Core circuits and wherein said AHaH circuit comprises an AHaH node circuit; and   at least one kT-RAM circuit that comprises at least one kT-Core circuit among said collection of said kT-Core circuits partitioned into AHaH nodes via temporal multiplexing.   
     
     
         27 . The circuit of  claim 26  wherein said AHaH node circuit provides a universal computational and adaptation resource and wherein said AHaH circuit is based on an AHaH node connection topology that provides for at least one of the following: classification, prediction, anomaly detection, clustering, feature learning, actuation, combinatorial optimization and universal logic. 
     
     
         28 . The circuit of  claim 26  wherein said AHaH circuit comprises at least one memristor.

Cited by (0)

No later patents cite this yet.

References (0)

No backward citations on record.