Selecting feature types to extract based on pre-classification of sensor measurements
First Claim
1. A method comprising:
- at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method;
receiving sensor measurements generated by one or more sensors of one or more devices;
pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications;
selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements, including;
in accordance with a determination that the sensor measurements correspond to a first pre-classification, selecting a first set of feature types as the one or more selected feature types; and
in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, selecting a second set of feature types as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types;
extracting features of the one or more selected feature types from the sensor measurements, including;
in accordance with a determination that the sensor measurements correspond to the first pre-classification, forgoing extracting features of one or more of the second set of feature types while extracting features of the first set of feature types; and
in accordance with a determination that the set of sensor measurements correspond to the second pre-classification, forgoing extracting features of one or more of the first set of feature types while extracting features of the second set of feature types; and
determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements.
3 Assignments
0 Petitions
Accused Products
Abstract
A processing apparatus including one or more processors and memory receives sensor measurements generated by one or more sensors of one or more devices, pre-classifies the sensor measurements as belonging to one of a plurality of pre-classifications, and selects one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements. The processing apparatus also extracts features of the one or more selected feature types from the sensor measurements and determines a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements.
-
Citations
159 Claims
-
1. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements, including; in accordance with a determination that the sensor measurements correspond to a first pre-classification, selecting a first set of feature types as the one or more selected feature types; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, selecting a second set of feature types as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types; extracting features of the one or more selected feature types from the sensor measurements, including; in accordance with a determination that the sensor measurements correspond to the first pre-classification, forgoing extracting features of one or more of the second set of feature types while extracting features of the first set of feature types; and in accordance with a determination that the set of sensor measurements correspond to the second pre-classification, forgoing extracting features of one or more of the first set of feature types while extracting features of the second set of feature types; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23)
-
24. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements, including; in accordance with a determination that the sensor measurements correspond to a first pre-classification, selecting a first set of feature types as the one or more selected feature types; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, selecting a second set of feature types as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types; extracting features of the one or more selected feature types from the sensor measurements, including; in accordance with a determination that the sensor measurements correspond to the first pre-classification, forgoing extracting features of one or more of the second set of feature types while extracting features of the first set of feature types; and in accordance with a determination that the set of sensor measurements correspond to the second pre-classification, forgoing extracting features of one or more of the first set of feature types while extracting features of the second set of feature types; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46)
-
-
47. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements, wherein the selecting includes; in accordance with a determination that the sensor measurements correspond to a first pre-classification, selecting a first set of feature types as the one or more selected feature types; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, selecting a second set of feature types as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types; extract features of the one or more selected feature types from the sensor measurements, wherein the extracting includes; in accordance with a determination that the sensor measurements correspond to the first pre-classification, forgoing extracting features of one or more of the second set of feature types while extracting features of the first set of feature types; and in accordance with a determination that the set of sensor measurements correspond to the second pre-classification, forgoing extracting features of one or more of the first set of feature types while extracting features of the second set of feature types; and determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69)
-
-
70. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes identifying the sensor measurements as belonging to one of a plurality of coupling pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes determining a coupling state of the respective device in accordance with a coupling classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein the coupling state of the respective device is selected from a set consisting of; a plurality of coupling-stable states corresponding to a coupling between the respective device and an entity associated with the respective device; and a plurality of coupling-transition states corresponding to a transition between two of the coupling-stable states. - View Dependent Claims (71, 72, 73, 74)
-
75. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes identifying the sensor measurements as belonging to one of a plurality of coupling pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes determining a coupling state of the respective device in accordance with a coupling classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein the coupling state of the respective device is selected from a set consisting of; a plurality of coupling-stable states corresponding to a coupling between the respective device and an entity associated with the respective device; and a plurality of coupling-transition states corresponding to a transition between two of the coupling-stable states. - View Dependent Claims (76, 77, 78, 79)
-
-
80. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes identifying the sensor measurements as belonging to one of a plurality of coupling pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes determining a coupling state of the respective device in accordance with a coupling classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein the coupling state of the respective device is selected from a set consisting of; a plurality of coupling-stable states corresponding to a coupling between the respective device and an entity associated with the respective device; and a plurality of coupling-transition states corresponding to a transition between two of the coupling-stable states. - View Dependent Claims (81, 82, 83, 84)
-
-
85. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes extracting features of one or more pre-classification feature types from the sensor measurements; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements, wherein extracting features of the pre-classification feature types from the sensor measurements is more resource efficient than extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (86, 87, 88, 89)
-
90. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes extracting features of one or more pre-classification feature types from the sensor measurements; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements, wherein extracting features of the pre-classification feature types from the sensor measurements is more resource efficient than extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (91, 92, 93, 94)
-
-
95. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications, wherein pre-classifying the sensor measurements includes extracting features of one or more pre-classification feature types from the sensor measurements; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extract features of the one or more selected feature types from the sensor measurements, wherein extracting features of the pre-classification feature types from the sensor measurements is more resource efficient than extracting features of the one or more selected feature types from the sensor measurements; and determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (96, 97, 98, 99)
-
-
100. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes updating a Markov model, wherein; the Markov model includes a plurality of model states corresponding to states of the respective device and a plurality of sets of transition probabilities between the plurality of model states; and updating the Markov model includes; in accordance with a determination that the sensor measurements correspond to a first pre-classification, updating one or more model states in the Markov model in accordance with a first set of transition probabilities; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, updating one or more model states in the Markov model in accordance with a second set of transition probabilities different from the first set of transition probabilities. - View Dependent Claims (101, 102, 103, 104)
-
105. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes updating a Markov model, wherein; the Markov model includes a plurality of model states corresponding to states of the respective device and a plurality of sets of transition probabilities between the plurality of model states; and updating the Markov model includes; in accordance with a determination that the sensor measurements correspond to a first pre-classification, updating one or more model states in the Markov model in accordance with a first set of transition probabilities; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, updating one or more model states in the Markov model in accordance with a second set of transition probabilities different from the first set of transition probabilities. - View Dependent Claims (106, 107, 108, 109)
-
-
110. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extract features of the one or more selected feature types from the sensor measurements; determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements, wherein determining the state of the respective device includes updating a Markov model, wherein; the Markov model includes a plurality of model states corresponding to states of the respective device and a plurality of sets of transition probabilities between the plurality of model states; and updating the Markov model includes; in accordance with a determination that the sensor measurements correspond to a first pre-classification, updating one or more model states in the Markov model in accordance with a first set of transition probabilities; and in accordance with a determination that the set of sensor measurements correspond to a second pre-classification, updating one or more model states in the Markov model in accordance with a second set of transition probabilities different from the first set of transition probabilities. - View Dependent Claims (111, 112, 113, 114)
-
-
115. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements; and wherein; the receiving sensor measurements, pre-classifying sensor measurements, selecting one or more feature types, extracting features and determining the state of the respective device are repeated for a plurality of measurement epochs that include a first epoch and a second epoch; during the first epoch; the sensor measurements are determined to correspond to a coupling-stable pre-classification; and a first set of feature types corresponding to the coupling-stable pre-classification are selected as the one or more selected feature types; and during the second epoch; the sensor measurements are determined to correspond to a coupling-transition pre-classification; and a second set of feature types corresponding to the coupling-transition pre-classification are selected as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types. - View Dependent Claims (116, 117, 118, 119)
-
120. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements; determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements; and wherein; the receiving sensor measurements, pre-classifying sensor measurements, selecting one or more feature types, extracting features and determining the state of the respective device are repeated for a plurality of measurement epochs that include a first epoch and a second epoch; during the first epoch; the sensor measurements are determined to correspond to a coupling-stable pre-classification; and a first set of feature types corresponding to the coupling-stable pre-classification are selected as the one or more selected feature types; and during the second epoch; the sensor measurements are determined to correspond to a coupling-transition pre-classification; and a second set of feature types corresponding to the coupling-transition pre-classification are selected as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types. - View Dependent Claims (121, 122, 123, 124)
-
-
125. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extract features of the one or more selected feature types from the sensor measurements; determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements; and wherein; the receiving sensor measurements, pre-classifying sensor measurements, selecting one or more feature types, extracting features and determining the state of the respective device are repeated for a plurality of measurement epochs that include a first epoch and a second epoch; during the first epoch; the sensor measurements are determined to correspond to a coupling-stable pre-classification; and a first set of feature types corresponding to the coupling-stable pre-classification are selected as the one or more selected feature types; and during the second epoch; the sensor measurements are determined to correspond to a coupling-transition pre-classification; and a second set of feature types corresponding to the coupling-transition pre-classification are selected as the one or more selected feature types, wherein the first set of feature types is different from the second set of feature types. - View Dependent Claims (126, 127, 128, 129)
-
-
130. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; scheduling extraction of features of a first subset of a plurality of feature types, wherein the first subset of feature types includes the one or more feature types selected based on the pre-classification; and scheduling extraction of features of a second subset of the plurality of feature types, wherein; the second subset of feature types includes feature types other than the one or more feature types selected based on the pre-classification; the extraction of features of the second subset of feature types is scheduled to occur after the extraction of features of the first subset of feature types; and the extraction of features of the second subset of feature types is subject to cancellation; extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (131, 132, 133, 134)
-
135. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; scheduling extraction of features of a first subset of a plurality of feature types, wherein the first subset of feature types includes the one or more feature types selected based on the pre-classification; and scheduling extraction of features of a second subset of the plurality of feature types, wherein; the second subset of feature types includes feature types other than the one or more feature types selected based on the pre-classification; the extraction of features of the second subset of feature types is scheduled to occur after the extraction of features of the first subset of feature types; and the extraction of features of the second subset of feature types is subject to cancellation; extracting features of the one or more selected feature types from the sensor measurements; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (136, 137, 138, 139)
-
-
140. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; schedule extraction of features of a first subset of a plurality of feature types, wherein the first subset of feature types includes the one or more feature types selected based on the pre-classification; and schedule extraction of features of a second subset of the plurality of feature types, wherein; the second subset of feature types includes feature types other than the one or more feature types selected based on the pre-classification; the extraction of features of the second subset of feature types is scheduled to occur after the extraction of features of the first subset of feature types; and the extraction of features of the second subset of feature types is subject to cancellation; extract features of the one or more selected feature types from the sensor measurements; and determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (141, 142, 143, 144)
-
-
145. A method comprising:
at a processing apparatus having one or more processors and memory storing one or more programs that, when executed by the one or more processors, cause the respective processing apparatus to perform the method; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements, wherein extracting the features of the one or more selected feature types includes extracting features of a first subset of a plurality of feature types, wherein the first subset includes the one or more feature types selected based on the pre-classification; and after extracting the features of the first subset; determining whether the features extracted from the first subset of feature types are consistent with the pre-classification; in accordance with a determination that the features extracted from the first subset of feature types are not consistent with the pre-classification, extracting features of a second subset of feature types that includes feature types other than the one or more feature types selected based on the pre-classification; and in accordance with a determination that the features extracted from the first subset of feature types are consistent with the pre-classification, forgoing extracting features of the second subset of feature types; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (146, 147, 148, 149)
-
150. A processing apparatus, comprising:
-
one or more processors; a set of one or more sensors; memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for; receiving sensor measurements generated by one or more sensors of one or more devices; pre-classifying the sensor measurements as belonging to one of a plurality of pre-classifications; selecting one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extracting features of the one or more selected feature types from the sensor measurements, wherein extracting the features of the one or more selected feature types includes extracting features of a first subset of a plurality of feature types, wherein the first subset includes the one or more feature types selected based on the pre-classification; and after extracting the features of the first subset; determining whether the features extracted from the first subset of feature types are consistent with the pre-classification; in accordance with a determination that the features extracted from the first subset of feature types are not consistent with the pre-classification, extracting features of a second subset of feature types that includes feature types other than the one or more feature types selected based on the pre-classification; and in accordance with a determination that the features extracted from the first subset of feature types are consistent with the pre-classification, forgoing extracting features of the second subset of feature types; and determining a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (151, 152, 153, 154)
-
-
155. A non-transitory computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a processing apparatus with one or more processors, cause the processing apparatus to:
-
receive sensor measurements generated by one or more sensors of one or more devices; pre-classify the sensor measurements as belonging to one of a plurality of pre-classifications; select one or more feature types to extract from the sensor measurements based at least in part on the pre-classification of the sensor measurements; extract features of the one or more selected feature types from the sensor measurements, wherein extracting the features of the one or more selected feature types includes extracting features of a first subset of a plurality of feature types, wherein the first subset includes the one or more feature types selected based on the pre-classification; and after extracting the features of the first subset; determine whether the features extracted from the first subset of feature types are consistent with the pre-classification; in accordance with a determination that the features extracted from the first subset of feature types are not consistent with the pre-classification, extract features of a second subset of feature types that includes feature types other than the one or more feature types selected based on the pre-classification; and in accordance with a determination that the features extracted from the first subset of feature types are consistent with the pre-classification, forgo extracting features of the second subset of feature types; and determine a state of a respective device of the one or more devices in accordance with a classification of the sensor measurements determined based on the one or more features extracted from the sensor measurements. - View Dependent Claims (156, 157, 158, 159)
-
Specification