CurriculumNet: Weakly Supervised Learning from Large-Scale Web Images

Guo, Sheng; Huang, Weilin; Zhang, Haozhi; Zhuang, Chijie; Dong, Dengke; Scott, Matthew R.; Huang, Ding-Long

doi:10.48550/arxiv.1808.01097

Cited by 8 publications

(7 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The idea of curriculum learning was originally proposed in [1], it demonstrates that the strategy of learning from easy to hard significantly improves the generalization of the deep model. Up to now, works been done via curriculum learning mainly focus on visual category discovery [29,41], object tracking [47], semi-/weaklysupervised learning [11,12,23,40], etc. [40] proposed an approach that processes multiple tasks in a sequence with sharing between subsequent tasks instead of solving all tasks jointly by finding the best order of tasks to be learned.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic Curriculum Learning for Imbalanced Data Classification

Wang

Gan

Yang

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

211

View full text Add to dashboard Cite

Human attribute analysis is a challenging task in the field of computer vision. One of the significant difficulties is brought from largely imbalance-distributed data. Conventional techniques such as re-sampling and cost-sensitive learning require prior-knowledge to train the system. To address this problem, we propose a unified framework called Dynamic Curriculum Learning (DCL) to adaptively adjust the sampling strategy and loss weight in each batch, which results in better ability of generalization and discrimination. Inspired by curriculum learning, DCL consists of two-level curriculum schedulers: (1) sampling scheduler which manages the data distribution not only from imbalance to balance but also from easy to hard; (2) loss scheduler which controls the learning importance between classification and metric learning loss. With these two schedulers, we achieve state-of-the-art performance on the widely used face attribute dataset CelebA and pedestrian attribute dataset RAP.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Very few works approach the imbalanced learning. Guo et al [12] developed a principled learning strategy by leveraging curriculum learning in a weakly supervised framework, with the goal of effectively learning from imbalanced data.…”

Section: Related Workmentioning

confidence: 99%

Dynamic Curriculum Learning for Imbalanced Data Classification

Wang

Gan

Yang

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

211

View full text Add to dashboard Cite

show abstract

“…With synthetic noisy labeled data, Rolnick et al [28] demonstrate that deep learning is robust to noise when training data is sufficiently large with large batch size and proper learning rate. Guo et al [6] develop a curriculum training scheme to learn noisy data from easy to hard. Jiang et al [10] design a MentorNet to adjust the loss weights of noisy samples in the training process.…”

Section: Noisy Data Learning Methodsmentioning

confidence: 99%

“…With the same full dataset, our guidance learning framework further improves the teacher model by 2.08%. As another useful trick in noisy data learning [25,6], fine-tuning the model trained with noisy data on clean set further boosts the final performance.This trick improves our guidance learning from 68.86% to 71.4%, and boosts the model#3 from 66.78% to 68.6% .…”

Section: Exploration Of Guidance Learning Onmentioning

confidence: 99%

“…With the same backbone, our guidance learning method improves the teacher model from 71.61% to 75.76% which outperforms [14] and [25]. CurriculumNet [6] uses the BNInception [9] network which is pre-trained on the full ImageNet with 21k classes as its backbone model, and is implemented with Caffe 1 . For a fair comparison, we replace the ResNet-50 as the same BNInception model, and obtain 78.77% without finetuning on clean set which is 2.93% better than CurriculumNet (78.77% vs. 75.8%).…”

Section: Experiments On Clothing1mmentioning

confidence: 99%

See 1 more Smart Citation

Product image recognition with guidance learning and noisy supervision

Peng

Cao

et al. 2020

Computer Vision and Image Understanding

View full text Add to dashboard Cite

This paper considers to recognize products from daily photos, which is an important problem in real-world applications but also challenging due to background clutters, category diversities, noisy labels, etc. We address this problem by two contributions. First, we introduce a novel largescale product image dataset, termed as Product-90. Instead of collecting product images by labor-and time-intensive image capturing, we take advantage of the web and download images from the reviews of several e-commerce websites where the images are casually captured by consumers. Labels are assigned automatically by the categories of ecommerce websites. Totally the Product-90 consists of more than 140K images with 90 categories. Due to the fact that consumers may upload unrelated images, it is inevitable that our Product-90 introduces noisy labels. As the second contribution, we develop a simple yet efficient guidance learning (GL) method for training convolutional neural networks (CNNs) with noisy supervision. The GL method first trains an initial teacher network with the full noisy dataset, and then trains a target/student network with both large-scale noisy set and small manually-verified clean set in a multi-task manner. Specifically, in the stage of student network training, the large-scale noisy data is supervised by its guidance knowledge which is the combination of its given noisy label and the soften label from the teacher network. We conduct extensive experiments on our Products-90 and public datasets, namely Food101, Food-101N, and Clothing1M. Our guidance learning method achieves performance superior to state-of-the-art methods on these datasets.

show abstract

Exploiting Curriculum Learning in Unsupervised Neural Machine Translation

Lu¹,

Zhang

2021

Findings of the Association for Computational Linguistics: EMNLP 2021

View full text Add to dashboard Cite

Back-translation (BT) has become one of the de facto components in unsupervised neural machine translation (UNMT), and it explicitly makes UNMT have translation ability. However, all the pseudo bi-texts generated by BT are treated equally as clean data during optimization without considering the quality diversity, leading to slow convergence and limited translation performance. To address this problem, we propose a curriculum learning method to gradually utilize pseudo bi-texts based on their quality from multiple granularities. Specifically, we first apply crosslingual word embedding to calculate the potential translation difficulty (quality) for the monolingual sentences. Then, the sentences are fed into UNMT from easy to hard batch by batch. Furthermore, considering the quality of sentences/tokens in a particular batch are also diverse, we further adopt the model itself to calculate the fine-grained quality scores, which are served as learning factors to balance the contributions of different parts when computing loss and encourage the UNMT model to focus on pseudo data with higher quality. Experimental results on WMT 14 En↔Fr, WMT 16 En↔De, WMT 16 En↔Ro, and LDC En↔Zh translation tasks demonstrate that the proposed method achieves consistent improvements with faster convergence speed. 1

show abstract

CurriculumNet: Weakly Supervised Learning from Large-Scale Web Images

Cited by 8 publications

References 15 publications

Dynamic Curriculum Learning for Imbalanced Data Classification

Dynamic Curriculum Learning for Imbalanced Data Classification

Product image recognition with guidance learning and noisy supervision

Exploiting Curriculum Learning in Unsupervised Neural Machine Translation

Contact Info

Product

Resources

About