Systems and methods for adaptive smart environment automation
View Patent ↗Several embodiments of systems and methods for adaptive smart environment automation are described herein. In one embodiment, a computer implemented method includes determining a plurality of sequence patterns of data points in a set of input data corresponding to a plurality of sensors in a space. The input data include a plurality of data points corresponding to each of the sensors, and the sequence patterns are at least partially discontinuous. The method also includes generating a plurality of statistical models based on the plurality of sequence patterns, and the individual statistical models corresponding to an activity of a user. The method further includes recognizing the activity of the user based on the statistical models and additional input data from the sensors.
1. A computer implemented method, comprising:
a processor collecting input data from a plurality of sensors in a space, the input data including a plurality of sequential data points corresponding to the individual sensors, wherein each data point represents a sensor event;
a processor analyzing the input data received from the plurality of sensors to determine a sequence pattern of data points from the input data, an instance of the sequence pattern of data points being at least partially discontinuous; and
a processor controlling a control element in the space based on the determined sequence pattern of data points.
2. The computer implemented method of claim 1 wherein:
the input data are first input data;
analyzing the input data includes:
calculating a frequency of each of the data points in the first input data;
generating second input data from the first input data by eliminating data points with a corresponding frequency less than a preselected threshold;
moving a data window across the second input data;
determining a sequence pattern based on data points in the data window as the data window moves across the second input data; and
correlating the determined sequence pattern to an activity of a user in the space; and
controlling the control element includes:
recognizing the activity of the user based on the determined sequence pattern; and
controlling the control element in the space based on the recognized activity of the user.
3. The computer implemented method of claim 1 wherein analyzing the input data includes:
moving a data window across the input data; and
determining a sequence pattern based on data points in the data window as the data window moves across the input data, the sequence pattern comprising all variations of a single sequence pattern that occur in the input data and satisfy the following condition:
D
L
(
D
)
(
D
L
(
a
)
+
D
L
(
D
❘
a
)
)
*
Γ
a
>
C
where DL is a description length of a corresponding argument, D is a data point, a is a sequence pattern, C is a minimum compression threshold, and Γ is a discontinuity factor.
4. The computer implemented method of claim 1 wherein analyzing the input data includes:
moving a data window across the input data, the data window having a length of data points;
determining a sequence pattern based on data points in the data window as the data window moves across the input data, the sequence pattern comprising all variations of a single sequence pattern that occur in the input data and satisfy the following condition:
D
L
(
D
)
(
D
L
(
a
)
+
D
L
(
D
❘
a
)
)
*
Γ
a
>
C
where DL is a description length of a corresponding argument, D is a data point, a is a sequence pattern, C is a minimum compression threshold, and Γ is a discontinuity factor; and
incrementing the length of the data window and repeating moving the data window across the input data and determining a sequence pattern until a preselected number of iterations are reached.
5. The computer implemented method of claim 1 wherein:
analyzing the input data includes determining a plurality of sequence patterns based on data points in the input data, the individual sequence patterns comprising all variations of a single sequence pattern that occur in the input data; and
the method further includes grouping the plurality of determined sequence patterns into a plurality of clusters based on a similarity between the sequence patterns.
6. The computer implemented method of claim 1 wherein:
analyzing the input data includes:
determining a plurality of sequence patterns based on data points in the input data, the individual sequence patterns comprising all variations of a single sequence pattern that occur in the input data and satisfy the following condition:
D
L
(
D
)
(
D
L
(
a
)
+
D
L
(
D
❘
a
)
)
*
Γ
a
>
C
where DL is a description length of a corresponding argument, D is a data point, a is a sequence pattern, C is a minimum compression threshold, and Γ is a discontinuity factor; and
the method further includes grouping the plurality of determined sequence patterns into a plurality of clusters based on a similarity between the sequence patterns.
7. The computer implemented method of claim 1 wherein:
analyzing the input data includes:
determining a plurality of sequence patterns based on data points in the input data; and
the method further includes:
computing an edit distance between two sequence patterns in the plurality of sequence patterns; and
grouping the two sequence patterns into a cluster if the computed edit distance is less than a threshold value.
8. The computer implemented method of claim 1 wherein:
analyzing the input data includes determining a plurality of sequence patterns based on data points in the input data, the individual sequence patterns comprising all variations of a single sequence pattern that occur in the input data; and
the method further includes:
grouping the plurality of determined sequence patterns into a plurality of clusters based on a similarity between the sequence patterns; and
generating a statistical model for the individual clusters, the statistical models including at least one of a Dynamic Bayes Network, a Naïve Bayes Classifier, a Markov model, and a hidden Markov model.
9. The computer implemented method of claim 1 wherein:
analyzing the input data includes:
determining a plurality of sequence patterns based on data points in the input data; and
the method further includes:
computing an edit distance between two sequence patterns in the plurality of sequence patterns;
grouping the plurality of determined sequence patterns into a plurality of clusters based on the computed edit distance; and
generating a statistical model for the individual clusters, the statistical models including a hidden Markov model.
10. The computer implemented method of claim 1 wherein:
analyzing the input data includes:
determining a plurality of sequence patterns based on data points in the input data; and
the method further includes:
computing an edit distance between two sequence patterns in the plurality of sequence patterns;
grouping the plurality of determined sequence patterns into a plurality of clusters based on the computed edit distance;
generating a statistical model for the individual clusters, the statistical models including a hidden Markov model;
collecting additional input data from the plurality of sensors; and
computing a statistical probability that the additional input data are correlating to an activity of a user in the space.
11. A computer implemented method, comprising:
a computing device determining a plurality of a sequence patterns of data points in a set of input data corresponding to a plurality of sensors in a space, the input data including a plurality of data points corresponding to each of the sensors, wherein each data point represents a sensor event and wherein at least one of the sequence patterns is at least partially discontinuous;
a computing device generating a plurality of statistical models based on the plurality of sequence patterns, the individual statistical models corresponding to an activity of a user; and
a computing device recognizing the activity of the user based on the statistical models and additional input data from the sensors.
12. The computer implemented method of claim 11 wherein:
determining a plurality of sequence patterns of data points includes determining a plurality of sequence patterns of data points in the input data, the individual sequence patterns comprising all variations of a single sequence pattern that occur in the input data; and
the method further includes grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have a similarity greater than a preselected threshold.
13. The computer implemented method of claim 11 wherein:
determining a plurality of sequence patterns of data points includes determining a plurality of sequence patterns of data points in the input data, the individual sequence patterns comprising all variations of a single sequence pattern that occur in the input data; and
the method further includes grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold.
14. The computer implemented method of claim 11 , further comprising:
grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold; and
generating a plurality of statistical models includes generating a plurality of statistical models individually based on each of the grouped clusters.
15. The computer implemented method of claim 11 , further comprising:
grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold; and
generating a plurality of statistical models includes generating a plurality of hidden Markov models individually based on each of the grouped clusters.
16. The computer implemented method of claim 11 , further comprising:
grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold; and
generating a plurality of statistical models includes generating a plurality of hidden Markov models individually based on each of the grouped clusters;
wherein recognizing the activity of the user includes:
collecting the additional input data from the sensors; and
computing a probability based on the additional input data from the sensors and the generated hidden Markov Models, the probability indicating a likelihood that the additional input data from the sensors correspond to an activity of a user.
17. The computer implemented method of claim 11 , further comprising:
grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold; and
generating a plurality of statistical models includes generating a plurality of hidden Markov models individually based on each of the grouped clusters;
wherein recognizing the activity of the user includes:
collecting the additional input data from the sensors;
computing a probability based on the additional input data from the sensors and the generated hidden Markov Models, the probability indicating a likelihood that the additional input data from the sensors correspond to an activity of a user; and
automating a control element in the space based on the computed probability.
18. The computer implemented method of claim 11 , further comprising:
grouping the plurality of sequence patterns into a plurality of clusters, the individual clusters containing sequence patterns that have an edit distance less than a preselected threshold; and
generating a plurality of statistical models includes generating a plurality of hidden Markov models individually based on each of the grouped clusters;
wherein recognizing the activity of the user includes:
collecting the additional input data from the sensors;
computing a probability based on the additional input data from the sensors and the generated hidden Markov Models, the probability indicating a likelihood that the additional input data from the sensors correspond to an activity of a user, the activity having at least one of a start time and a duration;
identifying a change in the start time and/or the duration of the activity; and
generating an alarm to the user about the change.
19. A computer system, comprising:
a plurality of sensors installed in a space, the sensors being configured to provide input data including a plurality of data points corresponding to each of the sensors, wherein each data point represents a sensor event; and
a controller operatively coupled to the sensors, the controller including:
an activity miner configured to analyze the input data from the plurality of sensors to determine a plurality of sequence patterns of data points, the sequence patterns being at least partially discontinuous; and
an activity model configured to generate a plurality of statistical models based on the plurality of sequence patterns, the individual statistical models corresponding to an activity of a user.
20. The computer system of claim 19 wherein the activity miner includes:
a discontinuous varied-order sequential module configured to determine a plurality of sequence patterns of data points, wherein at least one of the sequence patterns is discontinuous; and
a clustering module configured to group the plurality of sequence patterns into a set of clusters based on similarity of the plurality of sequence patterns.
21. The computer system of claim 19 wherein:
the activity miner includes (1) a discontinuous varied-order sequential module configured to determine a plurality of sequence patterns of data points, wherein at least one of the sequence patterns is discontinuous and (2) a clustering module configured to group the plurality of sequence patterns into a set of clusters based on similarity of the plurality of sequence patterns; and
the activity model includes a hidden Markov model having a plurality of hidden nodes related to the plurality of sequence patterns via a plurality of corresponding probabilities.