US 11,748,592 B1
Devices, systems, and methods for learning and using artificially intelligent interactive memories
Jasmin Cosic, Miami, FL (US)
Assigned to STORYFILE, INC., Pacific Palisades, CA (US)
Filed by Jasmin Cosic, Miami, FL (US)
Filed on Jan. 7, 2017, as Appl. No. 15/401,024.
Application 15/401,024 is a continuation of application No. 15/016,280, filed on Feb. 5, 2016, granted, now 9,582,762.
Int. Cl. G06N 3/006 (2023.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01)
CPC G06N 3/006 (2013.01) [G06N 5/04 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A system for learning artificially intelligent interactive memories, the system implemented at least in part on one or more computing devices, the system comprising:
one or more processor circuits;
a memory unit coupled to the one or more processor circuits;
a first picture-capturing device configured to capture a stream of digital pictures of a first conversation participant and a second picture-capturing device configured to capture a stream of digital pictures of a second conversation participant, the first and the second picture-capturing devices coupled to the one or more processor circuits;
a first sound-capturing device configured to capture a stream of digital sound samples of the first conversation participant and a second sound-capturing device configured to capture a stream of digital sound samples of the second conversation participant, the first and the second sound-capturing devices coupled to the one or more processor circuits, wherein the one or more processor circuits are configured to:
a first activity detector to detect the first conversation participant's first conversational activity from at least one of the stream of digital pictures of the first conversation participant or the stream of digital sound samples of the first conversation participant;
a second activity detector to detect the second conversation participant's first conversational activity from at least one of the stream of digital pictures of the second conversation participant or the stream of digital sound samples of the second conversation participant;
generate a first round of conversational exchange in a sequence including playback of a recording of the first conversation participant's first conversational activity and playback of a recording of the second conversation participant's first conversational activity in order of the sequence;
cause the memory unit to store the first round of conversational exchange, wherein the memory unit stores at least a plurality of rounds of conversational exchange;
applying the plurality of rounds of conversational exchange onto a neural network having a plurality of layers wherein each layer includes one or more rounds of the plurality of rounds of conversational exchange, wherein successive layers can be connected by a Connection, wherein a Connection may be associated with an occurrence count, a weight, or a parameter; and
following a path through the neural network to simulate a conversation.