Pattern learning system
View Patent ↗According to one embodiment, a pattern learning system includes a pattern learning tool that receives event messages in a sequential manner from multiple sensors and forms multiple sub-sequences that each includes a trigger event message, a consequence event message, and one or more intermediary event messages. The pattern learning tool then generates multiple graphs that each represents a sub-sequence of the plurality of event messages. The pattern learning tool then combines the graphs into a combined graph according to a type of each event message, and determines a causal sequence from the combined graph according to a heaviest weighted directed path from the trigger event message to the consequence event message.
1. A pattern learning system comprising:
a pattern learning tool comprising memory storing instructions executable on a computing system, the pattern learning tool operable to:
receive a plurality of sequential event messages from one or more sensors;
generate a plurality of graphs that each represents a sub-sequence of the plurality of event messages, each graph comprising a trigger event message, a consequence event message, and one or more intermediary event messages;
combine the plurality of graphs to form a combined graph according to a type of each event message; and
determine a causal sequence from the combined graph according to a heaviest weighted directed path from the trigger event message to the consequence event message.
2. The pattern learning system of claim 1 , wherein the trigger event message is known, the pattern learning tool operable to:
for each sub-sequence, select the consequence event message to be a specified period of time after the trigger event message.
3. The pattern learning system of claim 1 , wherein the consequence event message is known, the pattern learning tool operable to:
for each sub-sequence, select the trigger event message to be a specified period of time before the consequence event message.
4. The pattern learning system of claim 1 , wherein each of the plurality of graphs include a plurality of edges representing a possible causal path between subsequent event messages, the pattern learning tool operable to:
combine the plurality of edges according to the type of each message associated the plurality of edges; and
determine the causal sequence according to the heaviest weighted directed path of the combined edges from the trigger event message to the consequence event message.
5. The pattern learning system of claim 1 , wherein the pattern learning tool is operable to:
determine one or more noisy event messages from among the one or more intermediary event messages, the one or more noisy event messages comprising those intermediary event messages that do not lie along the determined causal sequence.
6. The pattern learning system of claim 5 , wherein the pattern learning tool is operable to determine the one or more noisy event messages according to a specified threshold level.
7. A pattern learning method comprising:
receiving a plurality of sequential event messages from one or more sensors;
generating a plurality of graphs that each represents a sub-sequence of the plurality of event messages, each graph comprising a trigger event message, a consequence event message, and one or more intermediary event messages;
combining the plurality of graphs to form a combined graph according to a type of each event message; and
determining a causal sequence from the combined graph according to a heaviest weighted directed path from the trigger event message to the consequence event message.
8. The pattern learning method of claim 7 , further comprising:
for each sub-sequence, select the consequence event message to be a specified period of time after the trigger event message that is known.
9. The pattern learning method of claim 7 , further comprising:
for each sub-sequence, select the trigger event message to be a specified period of time before the consequence event message that is known.
10. The pattern learning method of claim 7 , further comprising:
combining the plurality of edges according to the type of each message associated a plurality of edges representing a possible causal path between subsequent event messages; and
determining the causal sequence according to the heaviest weighted directed path of the combined edges from the trigger event message to the consequence event message.
11. The pattern learning method of claim 7 , further comprising:
determining one or more noisy event messages from among the one or more intermediary event messages, the one or more noisy event messages comprising those intermediary event messages that do not lie along the determined causal sequence.
12. The pattern learning method of claim 11 , further comprising determining the one or more noisy event messages according to a specified threshold level.
13. A non-transitory, computer-readable storage medium having computer-readable instructions stored thereon that, when executed by a processor, implement a method, the method comprising:
receiving a plurality of sequential event messages from one or more sensors;
generating a plurality of graphs that each represents a sub-sequence of the plurality of event messages, each graph comprising a trigger event message, a consequence event message, and one or more intermediary event messages;
combining the plurality of graphs to form a combined graph according to a type of each event message; and
determining a causal sequence from the combined graph according to a heaviest weighted directed path from the trigger event message to the consequence event message.
14. The computer-readable storage medium of claim 13 , wherein the method further comprises:
for each sub-sequence, selecting the consequence event message to be a specified period of time after the trigger event message that is known.
15. The computer-readable storage medium of claim 13 , wherein the method further comprises:
for each sub-sequence, selecting the trigger event message to be a specified period of time before the consequence event message that is known.
16. The computer-readable storage medium of claim 13 , further operable to:
combining the plurality of edges according to the type of each message associated a plurality of edges representing a possible causal path between subsequent event messages; and
determining the causal sequence according to the heaviest weighted directed path of the combined edges from the trigger event message to the consequence event message.
17. The computer-readable storage medium of claim 13 , wherein the method further comprises:
determining one or more noisy event messages from among the one or more intermediary event messages, the one or more noisy event messages comprising those intermediary event messages that do not lie along the determined causal sequence.
18. The computer-readable storage medium of claim 17 , to wherein the method further comprises determining the one or more noisy event messages according to a specified threshold level.