| IEICE Technical Committee Submission System Conference Paper's Information |
Online Proceedings [Sign in] Tech. Rep. Archives |
| Go Top Page | Go Previous | [Japanese] / [English] |
| Paper Abstract and Keywords | ||
| Presentation | 2022-12-22 15:30 [Short Paper] Semi supervised image classification using unreliable pseudo label Jihong Hu, Yinhao Li, Yen-Wei Chen (Ritsumeikan Univ.)  IBISML2022-47 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | Semi-supervised learning (SSL), which automatically annotates unlabeled data with pseudo labels during training, has achieved great success on the image classification task. However, the training process suffers from confirmation bias due to the low quality pseudo labels. To alleviate the confirmation bias, existing SSL methods use a high confidence threshold to select the data with reliable pseudo labels and ignore the data with unreliable pseudo labels for training. Thus, many training samples with unreliable pseudo labels cannot be fully used, especially the hard samples in complex datasets, causing low model performance and slow convergence speed. To address this issue, we introduce a novel pseudo label-based SSL approach with contrastive learning, which treats the samples with unreliable pseudo labels as the negative samples of the certain categories during training. Consequently, these samples with unreliable pseudo labels could be used effectively during training process especially in the early stage. In addition, we design a selection strategy based on the distribution of model prediction results to separate reliable and unreliable pseudo labels. The experiment results on CIFAR100 dataset demonstrate the effectiveness of the proposed method. | |
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | Semi supervised learning / contrastive learning / pseudo label / image classification / / / / | |
| Reference Info. | IEICE Tech. Rep., vol. 122, no. 325, IBISML2022-47, pp. 24-29, Dec. 2022. | |
| Paper # | IBISML2022-47 | |
| Date of Issue | 2022-12-15 (IBISML) | |
| ISSN | Online edition: ISSN 2432-6380 | |
| Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| Download PDF | IBISML2022-47 |
|
| Conference Information | |
| Committee | IBISML |
| Conference Date | 2022-12-22 - 2022-12-23 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Kyoto University |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Machine Learning, etc. |
| Paper Information | |
| Registration To | IBISML |
| Conference Code | 2022-12-IBISML |
| Language | English |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Semi supervised image classification using unreliable pseudo label |
| Sub Title (in English) | |
| Keyword(1) | Semi supervised learning |
| Keyword(2) | contrastive learning |
| Keyword(3) | pseudo label |
| Keyword(4) | image classification |
| Keyword(5) | |
| Keyword(6) | |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Jihong Hu |
| 1st Author's Affiliation | Ritsumeikan University (Ritsumeikan Univ.) |
| 2nd Author's Name | Yinhao Li |
| 2nd Author's Affiliation | Ritsumeikan University (Ritsumeikan Univ.) |
| 3rd Author's Name | Yen-Wei Chen |
| 3rd Author's Affiliation | Ritsumeikan University (Ritsumeikan Univ.) |
| 4th Author's Name | |
| 4th Author's Affiliation | () |
| 5th Author's Name | |
| 5th Author's Affiliation | () |
| 6th Author's Name | |
| 6th Author's Affiliation | () |
| 7th Author's Name | |
| 7th Author's Affiliation | () |
| 8th Author's Name | |
| 8th Author's Affiliation | () |
| 9th Author's Name | |
| 9th Author's Affiliation | () |
| 10th Author's Name | |
| 10th Author's Affiliation | () |
| 11th Author's Name | |
| 11th Author's Affiliation | () |
| 12th Author's Name | |
| 12th Author's Affiliation | () |
| 13th Author's Name | |
| 13th Author's Affiliation | () |
| 14th Author's Name | |
| 14th Author's Affiliation | () |
| 15th Author's Name | |
| 15th Author's Affiliation | () |
| 16th Author's Name | |
| 16th Author's Affiliation | () |
| 17th Author's Name | |
| 17th Author's Affiliation | () |
| 18th Author's Name | |
| 18th Author's Affiliation | () |
| Speaker | Author-1 |
| Date Time | 2022-12-22 15:30:00 |
| Presentation Time | 15 minutes |
| Registration for | IBISML |
| Paper # | IBISML2022-47 |
| Volume (vol) | vol.122 |
| Number (no) | no.325 |
| Page | pp.24-29 |
| #Pages | 6 |
| Date of Issue | 2022-12-15 (IBISML) |
[Return to Top Page]