Method, apparatus and computer program product for activity recognition
In accordance with an example embodiment a method, apparatus and computer program product are provided. The method comprises receiving, at an apparatus, a sampled data associated with an activity from one or more sensors wirelessly coupled to the apparatus. The sampled data is generated at the one or more sensors based on a compressive sampling of an activity data associated with the activity. The compressive sampling is performed based on a sampling information. The activity is classified based at least on the sampled data. An error associated with the classification of the activity is determined. The sampling information is updated or retained based on a comparison of the error with a threshold error. The updated sampling information is utilized for generating an updated sampled data. The updated sampled data facilitates in reclassification of the activity.
1. A wearable apparatus comprising:
at least one processor; and
at least one memory comprising computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause the wearable apparatus to at least:
receive, at the wearable apparatus, sampling information comprising a basis matrix and selection information from a second apparatus wirelessly coupled to the wearable apparatus, wherein the wearable apparatus is embodied by a wearable device configured to be attached to a user;
perform an initial compressive sampling of an activity data associated with an activity that comprises a human behavior in the form of an action or a movement by the user, wherein the compressive sampling is performed based at least on the sampling information and a sampling criteria for generating a sampled data associated with the activity, wherein the activity data is measured by one or more sensors of the wearable device, wherein the initial compressive sampling is carried out such that a basis size of the sampled data is within a predetermined range, said basis size being previously known by the wearable apparatus and shared with the second apparatus;
send the sampled data to the second apparatus for facilitating classification of the activity;
receive, in an instance in which an error satisfying a threshold error in the classification of the activity, updated selection information from the second apparatus;
perform, in response to receiving the updated selection information, a subsequent compressive sampling of the activity data based at least on the updated selection information and the sampling criteria to generate updated sampled data having an updated basis size greater than the basis size, wherein the updated basis size is determined dynamically based on said error in the classification of the activity; and
send the updated sampled data to the second apparatus for facilitating reclassification of the activity.
2. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to store the sampling information.
3. The wearable apparatus according to claim 1 , wherein the sampling information comprises a basis matrix comprising basis vectors and the selection information for selecting a basis size and a subset of basis vectors from the basis matrix.
4. The wearable apparatus according to claim 3 , wherein the updated selection information has an increased basis size associated with the subset of the basis matrix.
5. The wearable apparatus according to claim 3 , wherein the updated selection information has an increased number of basis vectors.
6. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to:
select the subset of the basis matrix based at least on the selection information.
7. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to:
send the sampled data to the second apparatus for facilitating classification of the activity until a frame of the sampled data is sent to the second apparatus.
8. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to:
generate, at the wearable apparatus, the basis matrix associated with a space; and
store the basis matrix in the at least one memory of the wearable apparatus.
9. The wearable apparatus according to claim 8 , wherein the selection information comprises instructions for selecting a subset of the basis matrix, and wherein the wearable apparatus is further caused, at least in part, to either:
select the subset of the basis matrix; or
cause the second apparatus to select the subset of the basis matrix.
10. The wearable apparatus according to claim 9 , wherein the instructions comprise one or more basis vectors and the basis size, the one or more basis vectors and the basis size operable to be selected from the basis matrix.
11. The wearable apparatus according to claim 10 , wherein one or more of the initial compressive sampling and the subsequent compressive sampling is performed utilizing an incoherent basis.
12. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to:
provide, based at least upon the reclassification of the activity by the user, activity-dependent user experiences.
13. The wearable apparatus according to claim 1 , wherein the activity is classified based at least on the sampled data and a classification rule.
14. The wearable apparatus according to claim 13 , wherein the classification rule comprises a model comprising a training set of labeled samples.
15. The wearable apparatus according to claim 14 , wherein the training set is represented in the basis matrix that is utilized for said sampling of the activity data or the training set is provided initially in a complete basis, said complete basis operable for calculating a set in any undercomplete basis used for compressive sampling of the activity data.
16. The wearable apparatus according to claim 13 , wherein labels associated with the training set are obtained using an unsupervised learning algorithm.
17. The wearable apparatus according to claim 13 , wherein labels associated with the training set are deduced from other sensory data received at or measured by the wearable apparatus or the second apparatus.
18. The wearable apparatus according to claim 1 , wherein the error is determined based upon one or more variances in an output of the classification of the activity.
19. The wearable apparatus according to claim 1 , wherein the wearable apparatus is further caused, at least in part, to:
cause the second apparatus to apply an error correction to the output of the classification.
20. The wearable apparatus according to claim 19 , wherein the error correction comprises a majority voting complex correlation approach.
21. The wearable apparatus according to claim 1 , wherein the updated selection information received from the second apparatus is operable for selecting at least a portion of the sampling information, wherein the at least the portion of the sampling information selected based on the updated selection information is different from the subset of sampling information utilized for generating the sampled data.
22. The wearable apparatus according to claim 21 , wherein the at least a portion of the sampling information selected based on the updated selection information is operable to be utilized for reclassifying the activity.