Machine learning algorithm trained to identify algorithmically populated shopping carts as candidates for verification

Patent metadata

Field	Value
Publication number	US-11475657-B2
Application number	US-202117174634-A
Country	US
Kind code	B2
Filing date	Feb 12, 2021
Priority date	Oct 25, 2019
Publication date	Oct 18, 2022
Grant date	Oct 18, 2022

Abstract

An apparatus includes a memory and processor. The memory stores a machine learning algorithm configured to decide whether to use a virtual shopping cart to verify all or a portion of a transaction performed with an algorithmic shopping cart. The processor receives feedback for a decision made by the algorithm, indicating whether the algorithmic and virtual carts match. The processor assigns a reward to the feedback. A first positive reward is assigned when the virtual shopping cart is not used for verification, and the feedback indicates that the carts match. A second positive reward is assigned when the virtual cart is used for verification, and the feedback indicates that the carts do not match. A first negative reward is assigned when the virtual shopping cart is not used for verification, and the feedback indicates that the carts do not match. A second negative reward is assigned when the virtual cart is used for verification, and the feedback indicates that the carts match. The processor uses the reward to update the algorithm.

First claim

What is claimed is:

1. An apparatus comprising:
a memory configured to store:
instructions corresponding to a machine learning algorithm configured, when implemented by one or more hardware processors, to use a set of inputs to determine whether to use a virtual shopping cart to verify all or a portion of a transaction performed with an algorithmic shopping cart, wherein:
the set of inputs comprises information collected from sensors located in a physical store during a shopping session of a person in the physical store;
the transaction is associated with the shopping session of the person;
the algorithmic shopping cart comprises a first set of items determined by an algorithm, based on the set of inputs, to have been selected by the person during the shopping session; and
the virtual shopping cart comprises a second set of items associated with the shopping session; and

the one or more hardware processors communicatively coupled to the memory, and configured to:
receive feedback for a decision made by the machine learning algorithm, wherein:
the decision made by the machine learning algorithm comprises at least one of:
a decision not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart; or
a decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart; and

the feedback indicates at least one of:
the algorithmic shopping cart matches the virtual shopping cart; or
the algorithmic shopping cart does not match the virtual shopping cart;

assign a reward value to the feedback, wherein the reward value comprises at least one of:
a first positive reward value, wherein the decision made by the machine learning algorithm comprises the decision not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart matches the virtual shopping cart;
a second positive reward value, wherein the decision made by the machine learning algorithm comprises the decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart does not match the virtual shopping cart;
a first negative reward value, wherein the decision made by the machine learning algorithm comprises the decision not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart does not match the virtual shopping cart; or
a second negative reward value, wherein the decision made by the machine learning algorithm comprises the decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart matches the virtual shopping cart; and

use the reward value to update the machine learning algorithm.

2. The apparatus of claim 1 , wherein the machine learning algorithm comprises a set of weights and updating the machine learning algorithm comprises adjusting values of the weights.

3. The apparatus of claim 2 , wherein the machine learning algorithm is double deep-Q network.

4. The apparatus of claim 1 , wherein the feedback comprises information received from the person.

5. The apparatus of claim 4 , wherein the information received from the person comprises a request for a refund of a price of an item changed to an account belonging to the person during processing of the transaction.

6. The apparatus of claim 1 , wherein the absolute value of the first negative reward value is larger than the absolute value of the second negative reward value.

7. The apparatus of claim 1 , wherein the sensors comprise:
an array of cameras; and
a set of weight sensors.

8. A method comprising:
receiving feedback for a decision made by a machine learning algorithm, wherein:
the machine learning algorithm is configured to use a set of inputs to determine whether to use a virtual shopping cart to verify all or a portion of a transaction performed with an algorithmic shopping cart, wherein:
the set of inputs comprises information collected from sensors located in a physical store during a shopping session of a person in the physical store;
the transaction is associated with the shopping session of the person;
the algorithmic shopping cart comprises a first set of items determined by an algorithm, based on the set of inputs, to have been selected by the person during the shopping session; and
the virtual shopping cart comprises a second set of items associated with the shopping session;

the decision made by the machine learning algorithm comprises at least one of:
a decision to not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart to process the transaction; or
a decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart; and

the feedback indicates at least one of:
the algorithmic shopping cart matches the virtual shopping cart; or
the algorithmic shopping cart does not match the virtual shopping cart;

assigning a reward value to the feedback, wherein the reward value comprises at least one of:
a first positive reward value, wherein the decision made by the machine learning algorithm comprises the decision not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart matches the virtual shopping cart;
a second positive reward value, wherein the decision made by the machine learning algorithm comprises the decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart does not match the virtual shopping cart;
a first negative reward value, wherein the decision made by the machine learning algorithm comprises the decision not to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart does not match the virtual shopping cart; or
a second negative reward value, wherein the decision made by the machine learning algorithm comprises the decision to use the virtual shopping cart to review all or a portion of the transaction performed by the algorithmic shopping cart and the feedback indicates that the algorithmic shopping cart matches the virtual shopping cart; and

using the reward value to update the machine learning algorithm.

9. The method of claim 8 , wherein the machine learning algorithm comprises a set of weights and updating the machine learning algorithm comprises adjusting values of the weights.

10. The method of claim 9 , wherein the machine learning algorithm is double deep-Q network.

11. The method of claim 8 , wherein the feedback comprises information received from the person.

12. The method of claim 11 , wherein the information received from the person comprises a request for a refund of a price of an item changed to an account belonging to the person during processing of the transaction.

13. The method of claim 8 , wherein the absolute value of the first negative reward value is larger than the absolute value of the second negative reward

Assignees

• 7 Eleven Inc

Inventors

• Wang Xinan
• Magee Matthew Raymond
• Dalal Kyle
• Mirza Shahmeer Ali
• Maung Crystal

Classifications

• H04N21/47815

  Electronic shopping (payment schemes, payment architectures or payment protocols for electronic shopping systems G06Q20/12) · CPC title

• G06V20/10 Primary

  Terrestrial scenes (scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59) · CPC title

• H04N21/21805

  enabling multiple viewpoints, e.g. using a plurality of cameras · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N20/00

  Machine learning · CPC title