In this paper, we present a study on learning visual recognition models from large scale noisy web data. We build a new database called WebVision, which contains more than $2.4$ million web images crawled from the Internet by using queries generated from the 1,000 semantic concepts of the benchmark ILSVRC 2012 dataset. Meta information along with those web images (e.g., title, description, tags, etc.) are also crawled. A validation set and test set containing human annotated images are also provided to facilitate algorithmic development. Based on our new database, we obtain a few interesting observations: 1) the noisy web images are sufficient for training a good deep CNN model for visual recognition; 2) the model learnt from our WebVision database exhibits comparable or even better generalization ability than the one trained from the ILSVRC 2012 dataset when being transferred to new datasets and tasks; 3) a domain adaptation issue (a.k.a., dataset bias) is observed, which means the dataset can be used as the largest benchmark dataset for visual domain adaptation. Our new WebVision database and relevant studies in this work would benefit the advance of learning state-of-the-art visual models with minimum supervision based on web data.
ETH Zurich（苏黎世联邦理工学院）；Google Research（谷歌研究院）；Carnegie Mellon University（卡内基梅隆大学）
We present the 2017 WebVision Challenge, a public image recognition challenge designed for deep learning based on web images without instance-level human annotation. Following the spirit of previous vision challenges, such as ILSVRC, Places2 and PASCAL VOC, which have played critical roles in the development of computer vision by contributing to the community with large scale annotated data for model designing and standardized benchmarking, we contribute with this challenge a large scale web images dataset, and a public competition with a workshop co-located with CVPR 2017. The WebVision datas…
In many real-world datasets, like WebVision, the performance of DNN based classifier is often limited by the noisy labeled data. To tackle this problem, some image related side information, such as captions and tags, often reveal underlying relationships across images. In this paper, we present an efficient weakly supervised learning by using a Side Information Network (SINet), which aims to effectively carry out a large scale classification with severely noisy labels. The proposed SINet consists of a visual prototype module and a noise weighting module. The visual prototype module is designed…
A recurring focus of the deep learning community is towards reducing the labeling effort. Data gathering and annotation using a search engine is a simple alternative to generating a fully human-annotated and human-gathered dataset. Although web crawling is very time efficient, some of the retrieved images are unavoidably noisy, i.e. incorrectly labeled. Designing robust algorithms for training on noisy data gathered from the web is an important research perspective that would render the building of datasets easier. In this paper we conduct a study to understand the type of label noise to expec…
Malong Technologies（码隆科技）；Shenzhen Malong Artificial Intelligence Research Center（深圳码隆人工智能研究中心）
We present a simple yet efficient approach capable of training deep neural networks on large-scale weakly-supervised web images, which are crawled raw from the Internet by using text queries, without any human annotation. We develop a principled learning strategy by leveraging curriculum learning, with the goal of handling a massive amount of noisy labels and data imbalance effectively. We design a new learning curriculum by measuring the complexity of data using its distribution density in a feature space, and rank the complexity in an unsupervised manner. This allows for an efficient impleme…
NAVER CLOVA；NAVER AI Lab（NAVER人工智能实验室）；GIST（光州科学技术院）；Seoul National University（首尔国立大学）；Yonsei University（延世大学）
Learning under a continuously changing data distribution with incorrect labels is a desirable real-world problem yet challenging. A large body of continual learning (CL) methods, however, assumes data streams with clean labels, and online learning scenarios under noisy data streams are yet underexplored. We consider a more practical CL task setup of an online learning from blurry data stream with corrupted labels, where existing CL methods struggle. To address the task, we first argue the importance of both diversity and purity of examples in the episodic memory of continual learning models. T…
Google Research（谷歌研究院）；University of Adelaide（阿德莱德大学）
Recent advances in deep learning have relied on large, labelled datasets to train high-capacity models. However, collecting large datasets in a time- and cost-efficient manner often results in label noise. We present a method for learning from noisy labels that leverages similarities between training examples in feature space, encouraging the prediction of each example to be similar to its nearest neighbours. Compared to training algorithms that use multiple models or distinct stages, our approach takes the form of a simple, additional regularization term. It can be interpreted as an inductive…
Many advances of deep learning techniques originate from the efforts of addressing the image classification task on large-scale datasets. However, the construction of such clean datasets is costly and time-consuming since the Internet is overwhelmed by noisy images with inadequate and inaccurate tags. In this paper, we propose a Ubiquitous Reweighting Network (URNet) that learns an image classification model from large-scale noisy data. By observing the web data, we find that there are five key challenges, i.e., imbalanced class sizes, high intra-classes diversity and inter-class similarity, i…
SenseTime Research（商汤科技研究院）；Rice University（莱斯大学）；The Chinese University of Hong Kong（香港中文大学）；Shanghai Jiao Tong University（上海交通大学）
Webly supervised learning becomes attractive recently for its efficiency in data expansion without expensive human labeling. However, adopting search queries or hashtags as web labels of images for training brings massive noise that degrades the performance of DNNs. Especially, due to the semantic confusion of query words, the images retrieved by one query may contain tremendous images belonging to other concepts. For example, searching `tiger cat' on Flickr will return a dominating number of tiger images rather than the cat images. These realistic noisy samples usually have clear visual seman…
SenseTime Research（商汤科技研究院）；Rice University（莱斯大学）；The University of Hong Kong（香港大学）；The Chinese University of Hong Kong（香港中文大学）
This paper focuses on webly supervised learning (WSL), where datasets are built by crawling samples from the Internet and directly using search queries as web labels. Although WSL benefits from fast and low-cost data collection, noises in web labels hinder better performance of the image classification model. To alleviate this problem, in recent works, self-label supervised loss $\mathcal{L}_s$ is utilized together with webly supervised loss $\mathcal{L}_w$. $\mathcal{L}_s$ relies on pseudo labels predicted by the model itself. Since the correctness of the web label or pseudo label is usually …
In this thesis we investigate the effect of using web images to build a large scale database to be used along a deep learning method for a classification task. We replicate the ImageNet large scale database (ILSVRC-2012) from images collected from the web using 4 different download strategies varying: the search engine, the query and the image resolution. As a deep learning method, we will choose the Convolutional Neural Network that was very successful with recognition tasks; the AlexNet.
Successful training of deep neural networks with noisy labels is an essential capability as most real-world datasets contain some amount of mislabeled data. Left unmitigated, label noise can sharply degrade typical supervised learning approaches. In this paper, we present robust temporal ensembling (RTE), which combines robust loss with semi-supervised regularization methods to achieve noise-robust learning. We demonstrate that RTE achieves state-of-the-art performance across the CIFAR-10, CIFAR-100, ImageNet, WebVision, and Food-101N datasets, while forgoing the recent trend of label filterin…
Drexel University（德雷塞尔大学）；CWI（荷兰国家数学与计算机科学研究中心）；University of Amsterdam（阿姆斯特丹大学）
Normalized web distance (NWD) is a similarity or normalized semantic distance based on the World Wide Web or another large electronic database, for instance Wikipedia, and a search engine that returns reliable aggregate page counts. For sets of search terms the NWD gives a common similarity (common semantics) on a scale from 0 (identical) to 1 (completely different). The NWD approximates the similarity of members of a set according to all (upper semi)computable properties. We develop the theory and give applications of classifying using Amazon, Wikipedia, and the NCBI website from the National…
RIKEN Center for Life Science Technologies (CLST)（日本理化学研究所生命科学技术中心）；RIKEN CLST-JEOL Collaboration Center（日本理化学研究所CLST-JEOL合作中心）；RIKEN Compass to Healthy Life Research Complex Program（日本理化学研究所健康生活研究综合体指南项目）；RIKEN BioResource Center (BRC)（日本理化学研究所生物资源中心）；Advanced Center for Computing and Communication (ACCC), RIKEN（日本理化学研究所高级计算与通信中心）
We introduce the RIKEN Microstructural Imaging Metadatabase, a semantic web-based imaging database in which image metadata are described using the Resource Description Framework (RDF) and detailed biological properties observed in the images can be represented as Linked Open Data. The metadata are used to develop a large-scale imaging viewer that provides a straightforward graphical user interface to visualise a large microstructural tiling image at the gigabyte level. We applied the database to accumulate comprehensive microstructural imaging data produced by automated scanning electron micro…
Massachusetts Institute of Technology（麻省理工学院）；Facebook AI（Facebook人工智能）
A plethora of recent work has shown that convolutional networks are not robust to adversarial images: images that are created by perturbing a sample from the data distribution as to maximize the loss on the perturbed example. In this work, we hypothesize that adversarial perturbations move the image away from the image manifold in the sense that there exists no physical process that could have produced the adversarial image. This hypothesis suggests that a successful defense mechanism against adversarial images should aim to project the images back onto the image manifold. We study such defens…
Large scale image classification datasets often contain noisy labels. We take a principled probabilistic approach to modelling input-dependent, also known as heteroscedastic, label noise in these datasets. We place a multivariate Normal distributed latent variable on the final hidden layer of a neural network classifier. The covariance matrix of this latent variable, models the aleatoric uncertainty due to label noise. We demonstrate that the learned covariance structure captures known sources of label noise between semantically similar and co-occurring classes. Compared to standard neural net…
National University of Singapore（新加坡国立大学）；Singapore University of Technology and Design（新加坡科技设计大学）；University of Electronic Science and Technology of China（电子科技大学）
With the proliferation of e-commerce websites and the ubiquitousness of smart phones, cross-domain image retrieval using images taken by smart phones as queries to search products on e-commerce websites is emerging as a popular application. One challenge of this task is to locate the attention of both the query and database images. In particular, database images, e.g. of fashion products, on e-commerce websites are typically displayed with other accessories, and the images taken by users contain noisy background and large variations in orientation and lighting. Consequently, their attention is…
Consistency regularization is a commonly-used technique for semi-supervised and self-supervised learning. It is an auxiliary objective function that encourages the prediction of the network to be similar in the vicinity of the observed training samples. Hendrycks et al. (2020) have recently shown such regularization naturally brings test-time robustness to corrupted data and helps with calibration. This paper empirically studies the relevance of consistency regularization for training-time robustness to noisy labels. First, we make two interesting and useful observations regarding the consiste…
Stanford University（斯坦福大学）；Facebook AI Research（Facebook人工智能研究院）；Google Brain（谷歌大脑）
Memorization in over-parameterized neural networks could severely hurt generalization in the presence of mislabeled examples. However, mislabeled examples are hard to avoid in extremely large datasets collected with weak supervision. We address this problem by reasoning counterfactually about the loss distribution of examples with uniform random labels had they were trained with the real examples, and use this information to remove noisy examples from the training set. First, we observe that examples with uniform random labels have higher losses when trained with stochastic gradient descent un…
