Fraud detection using multi-task learning and/or deep learning

A computer program product (CPP) comprising:

• a set of storage device(s);
• and
• computer code stored collectively in the set of storage device(s), with the computer code including data and instructions to cause a processor(s) set to perform at least the following operations:
• training a multi-task learning algorithm including an underlying learning structure and multi-task instructions, the multi-task learning algorithm being programmed and/or structured to receive new data sets representing financial transactions and to selectively provide likely-fraud determinations and likely-not-fraud determinations,
• receiving, by the multi-task learning algorithm, a first new data set representing a first financial transaction, and
• applying the multi-task learning algorithm to data of the first new data set to determine that the first financial transaction is likely-fraud;
• wherein the multi-task learning instructions:
• (i) solve multiple learning tasks in a temporally overlapping manner while exploiting commonalities and differences across tasks resulting in improved learning efficiency and prediction accuracy for the task-specific models, when compared to training multiple models separately, and (ii) improve generalization by using the domain information contained in the training signals of related tasks as an inductive bias by learning tasks in parallel while using a shared representation so that what is learned for each task can help other tasks be learned.

A computer system (CS) comprising:

• a processor(s) set;
• a set of storage device(s);
• and
• computer code stored collectively in the set of storage device(s), with the computer code including data and instructions to cause the processor(s) set to perform at least the following operations:
• training a multi-task learning algorithm including an underlying learning structure and multi-task instructions, the multi-task learning algorithm being programmed and/or structured to receive new data sets representing financial transactions and to selectively provide likely-fraud determinations and likely-not-fraud determinations,
• receiving, by the multi-task learning algorithm, a first new data set representing a first financial transaction, and
• applying the multi-task learning algorithm to data of the first new data set to determine that the first financial transaction is likely-fraud;
• wherein the multi-task learning instructions:
• (i) solve multiple learning tasks in a temporally overlapping manner while exploiting commonalities and differences across tasks resulting in improved learning efficiency and prediction accuracy for the task-specific models, when compared to training multiple models separately, and (ii) improve generalization by using the domain information contained in the training signals of related tasks as an inductive bias by learning tasks in parallel while using a shared representation so that what is learned for each task can help other tasks be learned.