US 12,217,011 B2
Generating adaptive textual explanations of output predicted by trained artificial-intelligence processes
Yaqiao Luo, Toronto (CA); Jesse Cole Cresswell, Toronto (CA); Kin Kwan Leung, Toronto (CA); Kai Wang, Hockessin, DE (US); Atiyeh Ashari Ghomi, Toronto (CA); Caitlin Messick, Chadds Ford, PA (US); Lu Shu, Mount Laurel, NJ (US); Barum Rho, Toronto (CA); Maksims Volkovs, Toronto (CA); and Paige Elyse Dickie, Toronto (CA)
Assigned to The Toronto-Dominion Bank, Toronto (CA)
Filed by The Toronto-Dominion Bank, Toronto (CA)
Filed on Nov. 23, 2021, as Appl. No. 17/533,358.
Claims priority of provisional application 63/252,496, filed on Oct. 5, 2021.
Prior Publication US 2023/0103753 A1, Apr. 6, 2023
Int. Cl. G06F 40/40 (2020.01)
CPC G06F 40/40 (2020.01) 22 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
a memory storing instructions;
a communications interface; and
at least one processor coupled to the memory and the communications interface, the at least one processor being configured to execute the instructions to:
train adaptively an artificial intelligence process based on a plurality of training datasets associated with a first temporal interval, and validate the trained artificial intelligence process based on a plurality of validation datasets associated with a second temporal interval, wherein the second temporal interval is subsequent to the first temporal interval;
receive an input dataset based on elements of first interaction data associated with a third temporal interval and a temporal identifier characterizing a fourth temporal interval, the fourth temporal interval being subsequent to the third temporal interval, and the input dataset comprising feature values of a plurality of input features corresponding to the first interaction data;
based on an application of the trained artificial intelligence process to the input dataset, generate output data representative of a predicted likelihood of an occurrence of an event during the fourth temporal interval;
train an explainability process based on the plurality of validation datasets;
provision pairs of the feature values and Shapley feature values to the trained explainability process, where each pair comprises a corresponding one of the feature values and a Shapley feature value associated with the corresponding feature value;
based on an application of the trained explainability process to the pairs of the feature values and Shapley feature values, generate a first element of textual content that characterizes an outcome associated with the predicted likelihood of the occurrence of the event, the first element of textual content being associated with the corresponding feature value; and
transmit a portion of the output data and the first element of textual content to a computing system via the communications interface, wherein the computing system is configured to generate or modify second interaction data based on the portion of the output data, and to provision notification data comprising the first element of textual content to a device associated with the first interaction data.