Imbalanced Classification via Feature Dictionary-Based Minority Oversampling

Park, Minho; Song, Hwa Jeon; Kang, Dong‐oh

doi:10.1109/access.2022.3161510

Cited by 5 publications

(8 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The image feature constitutes daily, gender, and embellishment features for each outfit and proceeds similarly to the text feature. These features are obtained by prelearning using deep fashion‐learning data and constructed data [31]. Figure 1 shows how IMG and TXT features are combined with

W_{emb}

and

h_{L}

in the early and late fusion, respectively.…”

Section: Our Methodsmentioning

confidence: 99%

Dialog‐based multi‐item recommendation using automatic evaluation

Chung,

Kim,

Yoo

et al. 2023

ETRI Journal

Self Cite

View full text Add to dashboard Cite

In this paper, we describe a neural network‐based application that recommends multiple items using dialog context input and simultaneously outputs a response sentence. Further, we describe a multi‐item recommendation by specifying it as a set of clothing recommendations. For this, a multimodal fusion approach that can process both cloth‐related text and images is required. We also examine achieving the requirements of downstream models using a pretrained language model. Moreover, we propose a gate‐based multimodal fusion and multiprompt learning based on a pretrained language model. Specifically, we propose an automatic evaluation technique to solve the one‐to‐many mapping problem of multi‐item recommendations. A fashion‐domain multimodal dataset based on Koreans is constructed and tested. Various experimental environment settings are verified using an automatic evaluation method. The results show that our proposed method can be used to obtain confidence scores for multi‐item recommendation results, which is different from traditional accuracy evaluation.

show abstract

W_{emb}

and

h_{L}

in the early and late fusion, respectively.…”

Section: Our Methodsmentioning

confidence: 99%

Dialog‐based multi‐item recommendation using automatic evaluation

Chung,

Kim,

Yoo

et al. 2023

ETRI Journal

Self Cite

View full text Add to dashboard Cite

show abstract

“…Data re-sampling solves the problem of long-tailed distribution image classification from the data level. Re-sampling is the most widely used method [ 4 , 5 ] in processing long-tailed distribution image classification in depth learning, mainly including over-sampling [ 6 , 7 ], under-sampling [ 8 , 9 , 10 ] and mixed sampling [ 11 , 12 ].…”

Section: Related Workmentioning

confidence: 99%

“…The oversampling method mainly reduces the imbalance between the head class and the tail class by increasing the number of samples of the tail class [ 6 , 7 ]. Inspired by this, Gupta et al, proposed the repeated factor sampling method [ 13 ], which performs a re-balancing operation on the training data by increasing the sampling frequency of the tail image.…”

Section: Related Workmentioning

confidence: 99%

“…According to the principle of the oversampling method, this method simply repeats the positive example, which will cause overemphasis on the positive example, and it is easy to over fit the positive example. In 2022, Park et al, proposed a oversampling method based on feature dictionary [ 6 ], and built a feature dictionary through a pre-trained feature extractor. The method of synthesizing samples based on feature dictionaries enriches the diversity of minority class data by fine-tuning classifiers.…”

Section: Related Workmentioning

confidence: 99%

“…However, these classic methods generally have poor results. For example, in the case of Oversampling of tail classes, it may lead to overfitting of tail classes [ 6 , 7 ], and if there are errors or noises in the samples of tail classes, oversampling may exacerbate these problems. Under-sampling may lead to insufficient learning of head classes [ 8 , 9 , 10 ] and may result in the loss of valuable data in the head classes.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

A Long-Tailed Image Classification Method Based on Enhanced Contrastive Visual Language

Song

Wang

2023

Sensors

View full text Add to dashboard Cite

To solve the problem that the common long-tailed classification method does not use the semantic features of the original label text of the image, and the difference between the classification accuracy of most classes and minority classes are large, the long-tailed image classification method based on enhanced contrast visual language trains the head class and tail class samples separately, uses text image to pre-train the information, and uses the enhanced momentum contrastive loss function and RandAugment enhancement to improve the learning of tail class samples. On the ImageNet-LT long-tailed dataset, the enhanced contrasting visual language-based long-tailed image classification method has improved all class accuracy, tail class accuracy, middle class accuracy, and the F1 value by 3.4%, 7.6%, 3.5%, and 11.2%, respectively, compared to the BALLAD method. The difference in accuracy between the head class and tail class is reduced by 1.6% compared to the BALLAD method. The results of three comparative experiments indicate that the long-tailed image classification method based on enhanced contrastive visual language has improved the performance of tail classes and reduced the accuracy difference between the majority and minority classes.

show abstract