On-manifold Adversarial Data Augmentation Improves Uncertainty Calibration

Patel, Kanil; Beluch, William; Zhang, Dan; Pfeiffer, Michael; Yang, Bin

doi:10.1109/icpr48806.2021.9413010

Cited by 18 publications

(13 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As has been empirically shown in [280] and [288], the adaptive binning calibration measures aECE and aSCE are more robust to the number of bins than the corresponding equal-width binning calibration measures ECE and SCE.…”

Section: B Evaluating Calibration Qualitymentioning

confidence: 82%

“…According to [278] [15], [68], [274] 5 [31], [275] 6 [272], [209], [273], [274], [16] 8 [277] 9 [268], [270] 10 [278], [279], [234], [280] 11 [269], [266], [11], [271], [279] Fig. 9: Visualization of the different types of uncertainty calibration methods presented in this paper.…”

Section: A Calibration Methodsmentioning

confidence: 99%

“…Inspired by the expected calibration error (ECE, see Section V-B) Naeini et al [283] measured the calibration performance on the unmixed samples for each batch b by the differentiable squared differences between the batch accuracy and the mean confidence on the batch samples. The total loss is given as a weighted combination of the original loss on mixed and unmixed samples and the calibration measure evaluated only on the unmixed samples: [280], Hendrycks et al [277] showed that exposing classifiers to out-of-distribution examples at training can help to improve the calibration.…”

Section: A Calibration Methodsmentioning

confidence: 99%

See 2 more Smart Citations

A Survey of Uncertainty in Deep Neural Networks

Gawlikowski¹,

Tassi²,

Ali³

et al. 2021

Preprint

163

158

View full text Add to dashboard Cite

and safety-critical real world applications remains limited. The main factors responsible for this limitation are• the lack of expressiveness and transparency of a deep neural network's inference model, which makes it difficult to trust their outcomes [2],• the inability to distinguish between in-domain and out-ofdomain samples [11], [12] and the sensitivity to domain shifts [13],• the inability to provide reliable uncertainty estimates for a deep neural network's decision [14] and frequently occurring overconfident predictions [15], [16],• the sensitivity to adversarial attacks that make deep neural networks vulnerable for sabotage [17], [18], [19].

show abstract

Section: B Evaluating Calibration Qualitymentioning

confidence: 82%

Section: A Calibration Methodsmentioning

confidence: 99%

Section: A Calibration Methodsmentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Uncertainty in Deep Neural Networks

Gawlikowski¹,

Tassi²,

Ali³

et al. 2021

Preprint

163

158

View full text Add to dashboard Cite

show abstract

“…For automotive applications this creates unacceptable latencies, thus more computationally efficient methods are needed. Sampling-free uncertainty estimation [21] and data augmentation [22], [23] are candidates. In comparison to the methods above that modify the training process, a simple approach that requires no retraining of the models, is post-hoc calibration [2].…”

Section: Related Workmentioning

confidence: 99%

Investigation of Uncertainty of Deep Learning-based Object Classification on Radar Spectra

Patel

Beluch

Rambach

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Deep learning (DL) has recently attracted increasing interest to improve object type classification for automotive radar. In addition to high accuracy, it is crucial for decision making in autonomous vehicles to evaluate the reliability of the predictions; however, decisions of DL networks are non-transparent. Current DL research has investigated how uncertainties of predictions can be quantified, and in this article, we evaluate the potential of these methods for safe, automotive radar perception. In particular we evaluate how uncertainty quantification can support radar perception under (1) domain shift, (2) corruptions of input signals, and (3) in the presence of unknown objects. We find that in agreement with phenomena observed in the literature, deep radar classifiers are overly confident, even in their wrong predictions. This raises concerns about the use of the confidence values for decision making under uncertainty, as the model fails to notify when it cannot handle an unknown situation. Accurate confidence values would allow optimal integration of multiple information sources, e.g. via sensor fusion. We show that by applying state-of-the-art post-hoc uncertainty calibration, the quality of confidence measures can be significantly improved, thereby partially resolving the over-confidence problem. Our investigation shows that further research into training and calibrating DL networks is necessary and offers great potential for safe automotive object classification with radar sensors.

show abstract

“…However, applications with limited complexity overhead and latency require sampling-free and single-model based calibration methods. Examples include modifying the training loss [14], scalable Gaussian processes [18], sampling-free uncertainty estimation [22], data augmentation [20,25,28,7] and ensemble distribution distillation [17]. In comparison, a simple approach that requires no retraining of the models is post-hoc calibration [5].…”

Section: Related Workmentioning

confidence: 99%

Multi-Class Uncertainty Calibration via Mutual Information Maximization-based Binning

Patel

Beluch

Yang

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Post-hoc calibration is a common approach for providing high-quality confidence estimates of deep neural network predictions. Recent work has shown that widely used scaling methods underestimate their calibration error, while alternative Histogram Binning (HB) methods with verifiable calibration performance often fail to preserve classification accuracy. In the case of multi-class calibration with a large number of classes K, HB also faces the issue of severe sample-inefficiency due to a large class imbalance resulting from the conversion into K one-vs-rest class-wise calibration problems. The goal of this paper is to resolve the identified issues of HB in order to provide verified and calibrated confidence estimates using only a small holdout calibration dataset for bin optimization while preserving multi-class ranking accuracy. From an information-theoretic perspective, we derive the I-Max concept for binning, which maximizes the mutual information between labels and binned (quantized) logits. This concept mitigates potential loss in ranking performance due to lossy quantization, and by disentangling the optimization of bin edges and representatives allows simultaneous improvement of ranking and calibration performance. In addition, we propose a shared class-wise (sCW) binning strategy that fits a single calibrator on the merged training sets of all K class-wise problems, yielding reliable estimates from a small calibration set. The combination of sCW and I-Max binning outperforms the state of the art calibration methods on various evaluation metrics across different benchmark datasets and models, even when using only a small set of calibration data, e.g. 1k samples for ImageNet.

show abstract

On-manifold Adversarial Data Augmentation Improves Uncertainty Calibration

Cited by 18 publications

References 8 publications

A Survey of Uncertainty in Deep Neural Networks

A Survey of Uncertainty in Deep Neural Networks

Investigation of Uncertainty of Deep Learning-based Object Classification on Radar Spectra

Multi-Class Uncertainty Calibration via Mutual Information Maximization-based Binning

Contact Info

Product

Resources

About