Evaluating Reliability in Medical DNNs: A Critical Analysis of Feature and Confidence-Based OOD Detection

• URL: http://arxiv.org/abs/2408.17337v1
• Date: Fri, 30 Aug 2024 15:02:22 GMT
• Title: Evaluating Reliability in Medical DNNs: A Critical Analysis of Feature and Confidence-Based OOD Detection
• Authors: Harry Anthony, Konstantinos Kamnitsas,
• Abstract summary: OOD detection methods can be categorised as confidence-based (using the model's output layer for OOD detection) or feature-based (not using the output layer) We show that OOD artefacts can boost a model's softmax confidence in its predictions, due to correlations in training data among other factors. We also show that feature-based methods typically perform worse at distinguishing between inputs that lead to correct and incorrect predictions.
• Score: 2.9049649065453336
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Reliable use of deep neural networks (DNNs) for medical image analysis requires methods to identify inputs that differ significantly from the training data, called out-of-distribution (OOD), to prevent erroneous predictions. OOD detection methods can be categorised as either confidence-based (using the model's output layer for OOD detection) or feature-based (not using the output layer). We created two new OOD benchmarks by dividing the D7P (dermatology) and BreastMNIST (ultrasound) datasets into subsets which either contain or don't contain an artefact (rulers or annotations respectively). Models were trained with artefact-free images, and images with the artefacts were used as OOD test sets. For each OOD image, we created a counterfactual by manually removing the artefact via image processing, to assess the artefact's impact on the model's predictions. We show that OOD artefacts can boost a model's softmax confidence in its predictions, due to correlations in training data among other factors. This contradicts the common assumption that OOD artefacts should lead to more uncertain outputs, an assumption on which most confidence-based methods rely. We use this to explain why feature-based methods (e.g. Mahalanobis score) typically have greater OOD detection performance than confidence-based methods (e.g. MCP). However, we also show that feature-based methods typically perform worse at distinguishing between inputs that lead to correct and incorrect predictions (for both OOD and ID data). Following from these insights, we argue that a combination of feature-based and confidence-based methods should be used within DNN pipelines to mitigate their respective weaknesses. These project's code and OOD benchmarks are available at: https://github.com/HarryAnthony/Evaluating_OOD_detection.

Related papers

• The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection [75.65876949930258]
  Out-of-distribution (OOD) detection is essential for model trustworthiness. We show that the superior OOD detection performance of state-of-the-art methods is achieved by secretly sacrificing the OOD generalization ability.
  arXiv  Detail & Related papers  (2024-10-12T07:02:04Z)
• Can OOD Object Detectors Learn from Foundation Models? [56.03404530594071]
  Out-of-distribution (OOD) object detection is a challenging task due to the absence of open-set OOD data. Inspired by recent advancements in text-to-image generative models, we study the potential of generative models trained on large-scale open-set data to synthesize OOD samples. We introduce SyncOOD, a simple data curation method that capitalizes on the capabilities of large foundation models.
  arXiv  Detail & Related papers  (2024-09-08T17:28:22Z)
• Can Pre-trained Networks Detect Familiar Out-of-Distribution Data? [37.36999826208225]
  We study the effect of PT-OOD on the OOD detection performance of pre-trained networks. We find that the low linear separability of PT-OOD in the feature space heavily degrades the PT-OOD detection performance. We propose a unique solution to large-scale pre-trained models: Leveraging powerful instance-by-instance discriminative representations of pre-trained models.
  arXiv  Detail & Related papers  (2023-10-02T02:01:00Z)
• Out-of-distribution Detection with Implicit Outlier Transformation [72.73711947366377]
  Outlier exposure (OE) is powerful in out-of-distribution (OOD) detection. We propose a novel OE-based approach that makes the model perform well for unseen OOD situations.
  arXiv  Detail & Related papers  (2023-03-09T04:36:38Z)
• Rethinking Out-of-distribution (OOD) Detection: Masked Image Modeling is All You Need [52.88953913542445]
  We find surprisingly that simply using reconstruction-based methods could boost the performance of OOD detection significantly. We take Masked Image Modeling as a pretext task for our OOD detection framework (MOOD)
  arXiv  Detail & Related papers  (2023-02-06T08:24:41Z)
• Augmenting Softmax Information for Selective Classification with Out-of-Distribution Data [7.221206118679026]
  We show that existing post-hoc methods perform quite differently compared to when evaluated only on OOD detection. We propose a novel method for SCOD, Softmax Information Retaining Combination (SIRC), that augments softmax-based confidence scores with feature-agnostic information. Experiments on a wide variety of ImageNet-scale datasets and convolutional neural network architectures show that SIRC is able to consistently match or outperform the baseline for SCOD.
  arXiv  Detail & Related papers  (2022-07-15T14:39:57Z)
• How Useful are Gradients for OOD Detection Really? [5.459639971144757]
  Out of distribution (OOD) detection is a critical challenge in deploying highly performant machine learning models in real-life applications. We provide an in-depth analysis and comparison of gradient based methods for OOD detection. We propose a general, non-gradient based method of OOD detection which improves over previous baselines in both performance and computational efficiency.
  arXiv  Detail & Related papers  (2022-05-20T21:10:05Z)
• DOODLER: Determining Out-Of-Distribution Likelihood from Encoder Reconstructions [6.577622354490276]
  This paper introduces and examines a novel methodology, DOODLER, for Out-Of-Distribution Detection. By training a Variational Auto-Encoder on the same data as another Deep Learning model, the VAE learns to accurately reconstruct In-Distribution (ID) inputs, but not to reconstruct OOD inputs. Unlike other work in the area, DOODLER requires only very weak assumptions about the existence of an OOD dataset, allowing for more realistic application.
  arXiv  Detail & Related papers  (2021-09-27T14:54:55Z)
• OODformer: Out-Of-Distribution Detection Transformer [15.17006322500865]
  In real-world safety-critical applications, it is important to be aware if a new data point is OOD. This paper proposes a first-of-its-kind OOD detection architecture named OODformer.
  arXiv  Detail & Related papers  (2021-07-19T15:46:38Z)
• Provably Robust Detection of Out-of-distribution Data (almost) for free [124.14121487542613]
  Deep neural networks are known to produce highly overconfident predictions on out-of-distribution (OOD) data. In this paper we propose a novel method where from first principles we combine a certifiable OOD detector with a standard classifier into an OOD aware classifier. In this way we achieve the best of two worlds: certifiably adversarially robust OOD detection, even for OOD samples close to the in-distribution, without loss in prediction accuracy and close to state-of-the-art OOD detection performance for non-manipulated OOD data.
  arXiv  Detail & Related papers  (2021-06-08T11:40:49Z)
• Robust Out-of-distribution Detection for Neural Networks [51.19164318924997]
  We show that existing detection mechanisms can be extremely brittle when evaluating on in-distribution and OOD inputs. We propose an effective algorithm called ALOE, which performs robust training by exposing the model to both adversarially crafted inlier and outlier examples.
  arXiv  Detail & Related papers  (2020-03-21T17:46:28Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.