公开(公告)号：US20240220809A1

公开(公告)日：2024-07-04

申请号：US18392227

申请日：2023-12-21

Applicant: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Changdong YOO ,  Haeyong KANG

IPC: G06N3/091 ,  G06N3/045

CPC classification number: G06N3/091 ,  G06N3/045

Abstract: A computing device performs a continual learning method of learning a plurality of task in a sequential order. The computing device uses, in a forward pass of a neural network for learning a current task of the plurality of tasks, a plurality of weights including selected weights, the selected weights being selected in a previous task of the plurality of tasks, freezes the selected weights and updates weights excluding the selected weights from the plurality of weights in a backward pass of the neural network for learning the current task, obtains a binary mask for selecting some weights of the plurality of weights based on a weight score of each of the plurality of weights, and finds a subnetwork of the neural network for the current task based on the binary mask.

公开(公告)号：US20240176819A1

公开(公告)日：2024-05-30

申请号：US18386311

申请日：2023-11-02

Applicant: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Changdong YOO ,  Sunjae YOON

IPC: G06F16/735 ,  G06F16/783 ,  G06V20/40

CPC classification number: G06F16/735 ,  G06F16/7837 ,  G06V20/41 ,  G06V20/46

Abstract: According to an embodiment of the present disclosure, a method for video moment retrieval performed by a computing device may include obtaining a pair of video and query to perform the video moment retrieval, determining based on meaning of the query whether a retrieval bias regarding the query has a positive effect on the video moment retrieval, and selectively removing the retrieval bias from a result of the video moment retrieval according to the determination result to generate a final retrieval result.

公开(公告)号：US20230316737A1

公开(公告)日：2023-10-05

申请号：US18023506

申请日：2020-09-15

Applicant: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Changdong YOO ,  Thang VU ,  Xuan Trung PHAM ,  Hyunjun JANG

IPC: G06V10/82 ,  G06V10/776

CPC classification number: G06V10/82 ,  G06V10/776

Abstract: The present invention provides a method comprising the steps of: generating a second anchor on the second convolutional feature map by scaling and shifting a first anchor in the ground-truth box; generating a third convolutional feature map by convolving the second convolutional feature map by means of a second convolution; determining whether the overlap ratio between the ground-truth box and a second single anchor is greater than or equal to a reference value; generating a third anchor by scaling and shifting the second anchor having an overlap ratio that is greater than or equal to the reference value; assigning an objectivity score to the third anchor; and presenting the third anchor, which has an objectivity score that is equal to or greater than a reference value, as a proposal on the third convolutional feature map.