Contextual application switch based on user behaviors

A method comprising:

• at a computing device comprising at least one processor, a communication interface, and a memory:
• capturing first edge frame of a first application, wherein the first application is being displayed at the computing device;
• applying one or more machine learning models to the first edge frame of the first application to identify a context tag for the first application;
• identifying a plurality of applications subsequently accessed in a sequential manner after accessing the first application in a context corresponding to the identified context tag, wherein each of the plurality of applications corresponds to one of a plurality of context level scores;
• identifying a second application of the plurality of applications, wherein the second application corresponds to a higher context level score than remaining applications of the plurality of applications;
• displaying, along with the first application, a selectable interface element corresponding to the second application of the plurality of applications, wherein selection of the selectable interface element causes the computing device to display a next application list that includes the second application;
• generating, prior to capturing the first edge frame of the first application, a plurality of training samples, wherein generating each of the plurality of training samples comprises:
• recording a timestamp corresponding to launch of an application;
• recording a timestamp corresponding to exit from the application;
• capturing, within a second predetermined period of time of the exit from the application, a corresponding edge frame of the application;
• and
• applying one or more machine learning algorithms to the corresponding edge frame to identify a corresponding context tag of the corresponding edge frame of the application, wherein applying the one or more machine learning algorithms comprises:
• comparing a usage frequency value of the application to a predetermined application popularity threshold;
• and
• in response to determining that the usage frequency value of the application exceeds the predetermined application popularity threshold:
• applying one or more supervised learning techniques to the corresponding edge frame of the application;
• identifying training samples for which the one or more supervised learning techniques were applied;
• displaying an authorization prompt requesting user input granting authorization for the identified training samples to be uploaded to a cloud computing device;
• uploading, in response to receiving the user input granting authorization, the identified training samples to the cloud computing device, wherein uploading the identified training samples causes the cloud computing device to update one or more stored image classifiers, each corresponding to a particular application;
• and
• downloading, once the one or more stored image classifiers have been updated based on the identified training samples, one or more updated image classifiers.

One or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, a communication interface, and memory, cause the computing platform to:

• capture first edge frame of a first application, wherein the first application is being displayed at the computing device;
• apply one or more machine learning models to the first edge frame of the first application to identify a context tag for the first application;
• identify a plurality of applications subsequently accessed in a sequential manner after accessing the first application in a context corresponding to the identified context, wherein each of the plurality of applications corresponds to one of a plurality of context level scores;
• identify a second application of the plurality of applications, wherein the second application corresponds to a higher context level score than remaining applications of the plurality of applications;
• display, along with the first application, a selectable interface element corresponding to the second application of the plurality of applications, wherein selection of the selectable interface element causes the computing device to display a next application list that includes the second application;
• generate, prior to capturing the first edge frame of the first application, a plurality of training samples, wherein generating each of the plurality of training samples comprises:
• recording a timestamp corresponding to launch of an application;
• recording a timestamp corresponding to exit from the application;
• capturing, within a second predetermined period of time of the exit from the application, a corresponding edge frame of the application;
• and
• applying one or more machine learning algorithms to the corresponding edge frame to identify a corresponding context tag of the corresponding edge frame of the application, wherein applying the one or more machine learning algorithms comprises:
• comparing a usage frequency value of the application to a predetermined application popularity threshold;
• and
• in response to determining that the usage frequency value of the application exceeds the predetermined application popularity threshold:
• applying one or more supervised learning techniques to the corresponding edge frame of the application;
• identify training samples for which the one or more supervised learning techniques were applied;
• display an authorization prompt requesting user input granting authorization for the identified training samples to be uploaded to a cloud computing device;
• upload, in response to receiving the user input granting authorization, the identified training samples to the cloud computing device, wherein uploading the identified training samples causes the cloud computing device to update one or more stored image classifiers, each corresponding to a particular application;
• and
• download, once the one or more stored image classifiers have been updated based on the identified training samples, one or more updated image classifiers.