ترغب بنشر مسار تعليمي؟ اضغط هنا

Combating Label Noise in Deep Learning Using Abstention

76   0   0.0 ( 0 )
 نشر من قبل Sunil Thulasidasan
 تاريخ النشر 2019
والبحث باللغة English




اسأل ChatGPT حول البحث

We introduce a novel method to combat label noise when training deep neural networks for classification. We propose a loss function that permits abstention during training thereby allowing the DNN to abstain on confusing samples while continuing to learn and improve classification performance on the non-abstained samples. We show how such a deep abstaining classifier (DAC) can be used for robust learning in the presence of different types of label noise. In the case of structured or systematic label noise -- where noisy training labels or confusing examples are correlated with underlying features of the data-- training with abstention enables representation learning for features that are associated with unreliable labels. In the case of unstructured (arbitrary) label noise, abstention during training enables the DAC to be used as an effective data cleaner by identifying samples that are likely to have label noise. We provide analytical results on the loss function behavior that enable dynamic adaption of abstention rates based on learning progress during training. We demonstrate the utility of the deep abstaining classifier for various image classification tasks under different types of label noise; in the case of arbitrary label noise, we show significant improvements over previously published results on multiple image benchmarks. Source code is available at https://github.com/thulas/dac-label-noise



قيم البحث

اقرأ أيضاً

Many Machine Learning algorithms, such as deep neural networks, have long been criticized for being black-boxes-a kind of models unable to provide how it arrive at a decision without further efforts to interpret. This problem has raised concerns on m odel applications trust, safety, nondiscrimination, and other ethical issues. In this paper, we discuss the machine learning interpretability of a real-world application, eXtreme Multi-label Learning (XML), which involves learning models from annotated data with many pre-defined labels. We propose a two-step XML approach that combines deep non-negative autoencoder with other multi-label classifiers to tackle different data applications with a large number of labels. Our experimental result shows that the proposed approach is able to cope with many-label problems as well as to provide interpretable label hierarchies and dependencies that helps us understand how the model recognizes the existences of objects in an image.
Label noise will degenerate the performance of deep learning algorithms because deep neural networks easily overfit label errors. Let X and Y denote the instance and clean label, respectively. When Y is a cause of X, according to which many datasets have been constructed, e.g., SVHN and CIFAR, the distributions of P(X) and P(Y|X) are entangled. This means that the unsupervised instances are helpful to learn the classifier and thus reduce the side effect of label noise. However, it remains elusive on how to exploit the causal information to handle the label noise problem. In this paper, by leveraging a structural causal model, we propose a novel generative approach for instance-dependent label-noise learning. In particular, we show that properly modeling the instances will contribute to the identifiability of the label noise transition matrix and thus lead to a better classifier. Empirically, our method outperforms all state-of-the-art methods on both synthetic and real-world label-noise datasets.
We study an extreme scenario in multi-label learning where each training instance is endowed with a single one-bit label out of multiple labels. We formulate this problem as a non-trivial special case of one-bit rank-one matrix sensing and develop an efficient non-convex algorithm based on alternating power iteration. The proposed algorithm is able to recover the underlying low-rank matrix model with linear convergence. For a rank-$k$ model with $d_1$ features and $d_2$ classes, the proposed algorithm achieves $O(epsilon)$ recovery error after retrieving $O(k^{1.5}d_1 d_2/epsilon)$ one-bit labels within $O(kd)$ memory. Our bound is nearly optimal in the order of $O(1/epsilon)$. This significantly improves the state-of-the-art sampling complexity of one-bit multi-label learning. We perform experiments to verify our theory and evaluate the performance of the proposed algorithm.
We propose a two-sample testing procedure based on learned deep neural network representations. To this end, we define two test statistics that perform an asymptotic location test on data samples mapped onto a hidden layer. The tests are consistent a nd asymptotically control the type-1 error rate. Their test statistics can be evaluated in linear time (in the sample size). Suitable data representations are obtained in a data-driven way, by solving a supervised or unsupervised transfer-learning task on an auxiliary (potentially distinct) data set. If no auxiliary data is available, we split the data into two chunks: one for learning representations and one for computing the test statistic. In experiments on audio samples, natural images and three-dimensional neuroimaging data our tests yield significant decreases in type-2 error rate (up to 35 percentage points) compared to state-of-the-art two-sample tests such as kernel-methods and classifier two-sample tests.
69 - Jie Shen , Chicheng Zhang 2020
This paper is concerned with computationally efficient learning of homogeneous sparse halfspaces in $mathbb{R}^d$ under noise. Though recent works have established attribute-efficient learning algorithms under various types of label noise (e.g. bound ed noise), it remains an open question when and how $s$-sparse halfspaces can be efficiently learned under the challenging malicious noise model, where an adversary may corrupt both the unlabeled examples and the labels. We answer this question in the affirmative by designing a computationally efficient active learning algorithm with near-optimal label complexity of $tilde{O}big({s log^4 frac d epsilon} big)$ and noise tolerance $eta = Omega(epsilon)$, where $epsilon in (0, 1)$ is the target error rate, under the assumption that the distribution over (uncorrupted) unlabeled examples is isotropic log-concave. Our algorithm can be straightforwardly tailored to the passive learning setting, and we show that the sample complexity is $tilde{O}big({frac 1 epsilon s^2 log^5 d} big)$ which also enjoys the attribute efficiency. Our main techniques include attribute-efficient paradigms for instance reweighting and for empirical risk minimization, and a new analysis of uniform concentration for unbounded data -- all of them crucially take the structure of the underlying halfspace into account.

الأسئلة المقترحة

التعليقات
جاري جلب التعليقات جاري جلب التعليقات
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا