US 12,169,784 B2
Online, incremental real-time learning for tagging and labeling data streams for deep neural networks and neural network applications
Lucas Neves, Somerville, MA (US); Liam Debeasi, Brookline, MA (US); Heather Ames Versace, Milton, MA (US); Jeremy Wurbs, Worcester, MA (US); Massimiliano Versace, Milton, MA (US); Warren Katz, Cambridge, MA (US); and Anatoli Gorchet, Newton, MA (US)
Assigned to Neurala, Inc., Boston, MA (US)
Filed by Neurala, Inc., Boston, MA (US)
Filed on Aug. 8, 2022, as Appl. No. 17/818,015.
Application 17/818,015 is a division of application No. 16/572,808, filed on Sep. 17, 2019, granted, now 11,410,033.
Application 16/572,808 is a continuation of application No. PCT/US2018/023155, filed on Mar. 19, 2018.
Claims priority of provisional application 62/472,925, filed on Mar. 17, 2017.
Prior Publication US 2022/0383115 A1, Dec. 1, 2022
Int. Cl. G06N 3/084 (2023.01); G06F 16/2455 (2019.01); G06F 16/58 (2019.01); G06F 17/15 (2006.01); G06F 18/24 (2023.01); G06N 3/08 (2023.01); G06V 10/44 (2022.01); G06V 10/74 (2022.01); G06V 10/764 (2022.01); G06V 30/19 (2022.01); G06V 30/194 (2022.01)
CPC G06N 3/084 (2013.01) [G06F 16/24568 (2019.01); G06F 16/5866 (2019.01); G06F 17/15 (2013.01); G06F 18/24 (2023.01); G06N 3/08 (2013.01); G06V 10/454 (2022.01); G06V 10/761 (2022.01); G06V 10/764 (2022.01); G06V 30/19133 (2022.01); G06V 30/19167 (2022.01); G06V 30/194 (2022.01)] 17 Claims
OG exemplary drawing
 
1. A method of tagging an object in a data stream, the method comprising:
extracting, with a neural network running on at least one processor, a first convolutional output from the data stream representing features of a first representation of an object in a first category;
classifying, with a classifier operably coupled to the neural network, the first representation of the object into the first category based on the first convolutional output;
tagging, by the at least one processor, the first representation of the object with a first tag;
displaying, via a user interface operably coupled to the classifier, the first tag and/or an indication of a position for the first representation of the object based on the first category;
performing an adjustment, by a user via the user interface, of the first tag and/or the indication of the position for the first representation of the object;
learning, by the classifier, the first tag and/or the indication of the position of the first representation of the object based on the adjustment without retraining the classifier on other categories of objects previously learned by the classifier; and
tagging, by the at least one processor based on the adjustment and learning, a second representation of an object in the first category present in the data stream along with the first representation of the object in the first category.