Dropbox, Inc.

Oz Rubin

Boris Borshevsky

Larry Xu

Gilad Gabriel Tsehori

Embodiments relate to improved classification techniques for classifying form (e.g., webpage) forms and inputs for password auto fill applications. A module may detect a client device accessing a form (e.g., webpage) and may identify inputs of the form (e.g., webpage). The module may group two or more of the identified inputs into a bundle and apply the bundle to one or more machine learned bundle classifier models configured to determine a form type of the bundle. The module may also apply the bundle to the selected input classifier model. The module may automatically fill one or more inputs of the form (e.g., webpage) according to the input types identified by the selected input classifier model.

Classifying form and inputs for password autofill

Embodiments relate to identifying an error in a form that is incorrectly filled out and apply corrections to the form. A module may identify inputs of a form (e.g., webpage) and apply an autofill template to the identified inputs, where the autofill template specifies input types of the identified inputs and content to be applied to the input types. The module may automatically detect an input error. The module may, responsive to detecting the input error, receive a correction indication specifying a first input of the form (e.g., webpage) and an input type of the first input. The module may update the autofill template according to the correction indication. The module may apply the updated autofill template to the first input of the form (e.g., webpage).

Growing community of inventors

Tel Aviv, Israel

Oz Rubin