FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2996297566> ?p ?o ?g. }

• W2996297566 abstract "Self-supervised learning approaches leverage unlabeled samples to acquire generic knowledge about different concepts, hence allowing for annotation-efficient downstream task learning. In this paper, we propose a novel self-supervised method that leverages multiple imaging modalities. We introduce the multimodal puzzle task, which facilitates rich representation learning from multiple image modalities. The learned representations allow for subsequent fine-tuning on different downstream tasks. To achieve that, we learn a modality-agnostic feature embedding by confusing image modalities at the data-level. Together with the Sinkhorn operator, with which we formulate the puzzle solving optimization as permutation matrix inference instead of classification, they allow for efficient solving of multimodal puzzles with varying levels of complexity. In addition, we also propose to utilize cross-modal generation techniques for multimodal data augmentation used for training self-supervised tasks. In other words, we exploit synthetic images for self-supervised pretraining, instead of downstream tasks directly, in order to circumvent quality issues associated with synthetic images, while improving data-efficiency and representations quality. Our experimental results, which assess the gains in downstream performance and data-efficiency, show that solving our multimodal puzzles yields better semantic representations, compared to treating each modality independently. Our results also highlight the benefits of exploiting synthetic images for self-supervised pretraining. We showcase our approach on four downstream tasks: Brain tumor segmentation and survival days prediction using four MRI modalities, Prostate segmentation using two MRI modalities, and Liver segmentation using unregistered CT and MRI modalities. We outperform many previous solutions, and achieve results competitive to state-of-the-art." @default.
• W2996297566 created "2019-12-26" @default.
• W2996297566 creator A5001459748 @default.
• W2996297566 creator A5023876634 @default.
• W2996297566 creator A5057136665 @default.
• W2996297566 creator A5074204533 @default.
• W2996297566 date "2019-12-11" @default.
• W2996297566 modified "2023-09-27" @default.
• W2996297566 title "Multimodal Self-Supervised Learning for Medical Image Analysis" @default.
• W2996297566 cites W1514535095 @default.
• W2996297566 cites W1641498739 @default.
• W2996297566 cites W165381567 @default.
• W2996297566 cites W1665115054 @default.
• W2996297566 cites W1901129140 @default.
• W2996297566 cites W1926131438 @default.
• W2996297566 cites W1933349210 @default.
• W2996297566 cites W1990283121 @default.
• W2996297566 cites W1997147102 @default.
• W2996297566 cites W2108598243 @default.
• W2996297566 cites W2128745377 @default.
• W2996297566 cites W2148300333 @default.
• W2996297566 cites W2156303437 @default.
• W2996297566 cites W2163605009 @default.
• W2996297566 cites W2184188583 @default.
• W2996297566 cites W219767539 @default.
• W2996297566 cites W2321533354 @default.
• W2996297566 cites W2326925005 @default.
• W2996297566 cites W2331128040 @default.
• W2996297566 cites W2339754110 @default.
• W2996297566 cites W2405756170 @default.
• W2996297566 cites W2461705298 @default.
• W2996297566 cites W2522453719 @default.
• W2996297566 cites W2546190447 @default.
• W2996297566 cites W2556930864 @default.
• W2996297566 cites W2605287558 @default.
• W2996297566 cites W2606014079 @default.
• W2996297566 cites W2608015370 @default.
• W2996297566 cites W2616788541 @default.
• W2996297566 cites W2619383789 @default.
• W2996297566 cites W2619697695 @default.
• W2996297566 cites W2628684354 @default.
• W2996297566 cites W2739192055 @default.
• W2996297566 cites W2742126485 @default.
• W2996297566 cites W2745006834 @default.
• W2996297566 cites W2751069891 @default.
• W2996297566 cites W2798976889 @default.
• W2996297566 cites W2802798675 @default.
• W2996297566 cites W2808604288 @default.
• W2996297566 cites W2842511635 @default.
• W2996297566 cites W2883190781 @default.
• W2996297566 cites W2883725317 @default.
• W2996297566 cites W2886300652 @default.
• W2996297566 cites W2889056793 @default.
• W2996297566 cites W2889779108 @default.
• W2996297566 cites W2892032594 @default.
• W2996297566 cites W2899699486 @default.
• W2996297566 cites W2915126261 @default.
• W2996297566 cites W2937845726 @default.
• W2996297566 cites W2944828972 @default.
• W2996297566 cites W2950178297 @default.
• W2996297566 cites W2950187998 @default.
• W2996297566 cites W2950577311 @default.
• W2996297566 cites W2950761309 @default.
• W2996297566 cites W2958360136 @default.
• W2996297566 cites W2962742544 @default.
• W2996297566 cites W2962756039 @default.
• W2996297566 cites W2962770129 @default.
• W2996297566 cites W2962825119 @default.
• W2996297566 cites W2962931121 @default.
• W2996297566 cites W2962960500 @default.
• W2996297566 cites W2962969419 @default.
• W2996297566 cites W2963073614 @default.
• W2996297566 cites W2963549470 @default.
• W2996297566 cites W2963663752 @default.
• W2996297566 cites W2963717741 @default.
• W2996297566 cites W2963758027 @default.
• W2996297566 cites W2963863924 @default.
• W2996297566 cites W2963904328 @default.
• W2996297566 cites W2964121744 @default.
• W2996297566 cites W2964316651 @default.
• W2996297566 cites W2975758481 @default.
• W2996297566 cites W2978777215 @default.
• W2996297566 cites W2979708377 @default.
• W2996297566 cites W2979888373 @default.
• W2996297566 cites W2981448404 @default.
• W2996297566 cites W2982151492 @default.
• W2996297566 cites W2984306354 @default.
• W2996297566 cites W2990873191 @default.
• W2996297566 cites W3005185514 @default.
• W2996297566 cites W3005680577 @default.
• W2996297566 cites W3014795415 @default.
• W2996297566 cites W3024463470 @default.
• W2996297566 cites W3025966429 @default.
• W2996297566 cites W3027445961 @default.
• W2996297566 cites W3034241283 @default.
• W2996297566 cites W3082337820 @default.
• W2996297566 cites W3100622512 @default.
• W2996297566 cites W3105747145 @default.
• W2996297566 cites W3106286734 @default.
• W2996297566 cites W3208720418 @default.

• next