Device operation based on dynamic classifier

A method comprising:

• receiving, at a processor, plurality of audio data samples corresponding to a plurality of different users;
• providing, at the processor, the plurality of audio data samples to a dynamic classifier,
• generating classification outputs, by the dynamic classifier integrated into the processor, corresponding to the different users based on unsupervised real time adaptive clustering of feature data of the audio data samples to adjust a decision boundary, of the dynamic classifier, between discriminative categories of the feature data space to distinguish between feature sets corresponding to the different users;
• performing an open-discourse authentication, by the dynamic classifier integrated into the processor, to on the fly enroll and authenticate each of the different users to the device such that one or more of the different users were initially non-authenticated, and after generation of the classification outputs the one or more of the different users are authenticated based on the classification outputs;
• and
• selectively accessing a particular device based on the classification outputs.

A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to:

• receive plurality of audio data samples corresponding to a plurality of different users;
• provide the plurality of audio data samples to a dynamic classifier, the dynamic classifier configured to:
• generate a classification outputs corresponding to the different users based on unsupervised real time adaptive clustering of feature data of the audio data samples to adjust a decision boundary, of the dynamic classifier, between discriminative categories of the feature data space to distinguish between feature sets corresponding to the different users;
• and
• perform an open-discourse authentication to on the fly enroll and authenticate each of the different users to the device such that one or more of the different users were initially non-authenticated, and after generation of the classification outputs the one or more of the different users are authenticated based on the classification outputs;
• and
• selectively access a particular device based on the classification outputs.

An apparatus comprising:

• means for receiving plurality of audio data samples corresponding to a plurality of different users;
• means for providing the plurality of audio data samples to a dynamic classifier, the dynamic classifier comprising:
• means for generating a classification outputs corresponding to the different users based on unsupervised real time adaptive clustering of feature data of the audio data samples to adjust a decision boundary, of the dynamic classifier, between discriminative categories of the feature data space to distinguish between feature sets corresponding to the different users;
• and
• means for performing an open-discourse authentication to on the fly enroll and authenticate each of the different users to the device such that one or more of the different users were initially non-authenticated, and after generation of the classification outputs the one or more of the different users are authenticated based on the classification outputs;
• and
• means for selectively accessing a particular device based on the classification outputs.