Now showing 1 - 10 of 44
PublicationConcept Correlation and its Effects on Concept-Based Models( 2023)
; ;Monnet, MaureenConcept-based learning approaches for image classification, such as Concept Bottleneck Models, aim to enable interpretation and increase robustness by directly learning high-level concepts which are used for predicting the main class. They achieve competitive test accuracies compared to standard end-to-end models. However, with multiple concepts per image and binary concept annotations (without concept localization), it is not evident if the output of the concept model is truly based on the predicted concepts or other features in the image. Additionally, high correlations between concepts would allow a model to predict a concept with high test accuracy by simply using a correlated concept as a proxy. In this paper, we analyze these correlations between concepts in the CUB and GTSRB datasets and propose methods beyond test accuracy for evaluating their effects on the performance of a concept-based model trained on this data. To this end, we also perform a more detailed analysis on the effects of concept correlation using synthetically generated datasets of 3D shapes. We see that high concept correlation increases the risk of a model's inability to distinguish these concepts. Yet simple techniques, like loss weighting, show promising initial results for mitigating this issue.
PublicationTowards Human-Interpretable Prototypes for Visual Assessment of Image Classification ModelsExplaining black-box Artificial Intelligence (AI) models is a cornerstone for trustworthy AI and a prerequisite for its use in safety critical applications such that AI models can reliably assist humans in critical decisions. However, instead of trying to explain our models post-hoc, we need models which are interpretable-by-design built on a reasoning process similar to humans that exploits meaningful high-level concepts such as shapes, texture or object parts. Learning such concepts is often hindered by its need for explicit specification and annotation up front. Instead, prototype-based learning approaches such as ProtoPNet claim to discover visually meaningful prototypes in an unsupervised way. In this work, we propose a set of properties that those prototypes have to fulfill to enable human analysis, e.g. as part of a reliable model assessment case, and analyse such existing methods in the light of these properties. Given a ‘Guess who?’ game, we find that these prototypes still have a long way ahead towards definite explanations. We quantitatively validate our findings by conducting a user study indicating that many of the learnt prototypes are not considered useful towards human understanding. We discuss about the missing links in the existing methods and present a potential real-world application motivating the need to progress towards truly human-interpretable prototypes.
PublicationTowards Human-Interpretable Prototypes for Visual Assessment of Image Classification ModelsExplaining black-box Artificial Intelligence (AI) models is a cornerstone for trustworthy AI and a prerequisite for its use in safety critical applications such that AI models can reliably assist humans in critical decisions. However, instead of trying to explain our models post-hoc, we need models which are interpretable-by-design built on a reasoning process similar to humans that exploits meaningful high-level concepts such as shapes, texture or object parts. Learning such concepts is often hindered by its need for explicit specification and annotation up front. Instead, prototype-based learning approaches such as ProtoPNet claim to discover visually meaningful prototypes in an unsupervised way. In this work, we propose a set of properties that those prototypes have to fulfill to enable human analysis, e.g. as part of a reliable model assessment case, and analyse such existing methods in the light of these properties. Given a 'Guess who?' game, we find that these prototypes still have a long way ahead towards definite explanations. We quantitatively validate our findings by conducting a user study indicating that many of the learnt prototypes are not considered useful towards human understanding. We discuss about the missing links in the existing methods and present a potential real-world application motivating the need to progress towards truly human-interpretable prototypes.
PublicationSEC-Learn: Sensor Edge Cloud for Federated Learning( 2022-04-27)
; ;Antes, Christoph ; ; ;Johnson, David S. ;Jung, Matthias ; ; ; ; ;Kutter, Christoph ; ;Loroch, Dominik M. ; ;Laleni, Nelli ; ;Leugering, Johannes ;Martín Fernández, Rodrigo ;Mateu, Loreto ;Mojumder, Shaown ; ; ; ; ; ;Wallbott, Paul ;Due to the slow-down of Moore’s Law and Dennard Scaling, new disruptive computer architectures are mandatory. One such new approach is Neuromorphic Computing, which is inspired by the functionality of the human brain. In this position paper, we present the projected SEC-Learn ecosystem, which combines neuromorphic embedded architectures with Federated Learning in the cloud, and performance with data protection and energy efficiency.
PublicationEnsemble-based Uncertainty Estimation with overlapping alternative Predictions( 2022)
; ;Schmoeller da Roza, FelippeA reinforcement learning model will predict an action in whatever state it is. Even if there is no distinct outcome due to unseen states the model may not indicate that. Methods for uncertainty estimation can be used to indicate this. Although a known approach in Machine Learning, most of the available uncertainty estimation methods are not able to deal with the choice overlap that happens in states where multiple actions can be taken by a reinforcement learning agent with a similar performance outcome. In this work, we investigate uncertainty estimation on simplified scenarios in a gridworld environment. Using ensemble-based uncertainty estimation we propose an algorithm based on action count variance (ACV) to deal with discrete action spaces and a calculation based on the in-distribution delta (IDD) of the action count variance to handle overlapping alternative predictions. To visualize the expressiveness of the model uncertainty we create heatmaps for different in-distribution (ID) and out-of-distribution (OOD) scenarios and propose an indicator for uncertainty. We can show that the method is able to indicate potentially unsafe states when the agent is facing novel elements in the OOD scenarios while capable to distinguish uncertainty resulting from OOD instances from uncertainty caused by the overlapping of alternative predictions.
PublicationIs it all a cluster game?( 2022)
; ;Koner, Rajat ;Günnemann, StephanIt is essential for safety-critical applications of deep neural networks to determine when new inputs are significantly different from the training distribution. In this paper, we explore this out-of-distribution (OOD) detection problem for image classification using clusters of semantically similar embeddings of the training data and exploit the differences in distance relationships to these clusters between in- and out-of-distribution data. We study the structure and separation of clusters in the embedding space and find that the supervised contrastive learning leads to well separated clusters while its self-supervised counterpart fails to do so. In our extensive analysis of different training methods, clustering strategies, distance metrics and thresholding approaches, we observe that there is no clear winner. The optimal approach depends on the model architecture and selected datasets for in- and out-of-distribution. While we could reproduce the outstanding results for contrastive training on CIFAR-10 as in-distribution data, we find standard cross-entropy paired with cosine similarity outperforms all contrastive training methods when training on CIFAR-100 instead. Cross-entropy provides competitive results as compared to expensive contrastive training methods.
PublicationBeyond Test Accuracy: The Effects of Model Compression on CNNs( 2022)
; ;Schwienbacher, KristianModel compression is widely employed to deploy convolutional neural networks on devices with limited computational resources or power limitations. For high stakes applications, such as autonomous driving, it is, however, important that compression techniques do not impair the safety of the system. In this paper, we therefore investigate the changes introduced by three compression methods - post-training quantization, global unstructured pruning, and the combination of both - that go beyond the test accuracy. To this end, we trained three image classifiers on two datasets and compared them regarding their performance on the class level and regarding their attention to different input regions. Although the deviations in test accuracy were minimal, our results show that the considered compression techniques introduce substantial changes to the models that reflect in the quality of predictions of individual classes and in the salience of input regions. While we did not observe the introduction of systematic errors or biases towards certain classes, these changes can significantly impact the failure modes of CNNs and thus are highly relevant for safety analyses. We therefore conclude that it is important to be aware of the changes caused by model compression and to already consider them in the early stages of the development process.
PublicationAI in MedTech Production. Visual Inspection for Quality Assurance( 2021)Automated visual inspection based on machine learning and computer vision algorithms is a promising approach to ensure the quality of critical medical implants and equipments. However, limited availability of data and potentially unpredictable deep learning models pose major challenges to bring such solutions to life and to the market. This talk addresses the open challenges as well as current research directions for dependable visual inspection in quality assurance of medical products.
PublicationOODformer: Out-Of-Distribution Detection Transformer( 2021)
;Koner, Rajat ; ; ;Günnemann, StephanTresp, VolkerA serious problem in image classification is that a trained model might perform well for input data that originates from the same distribution as the data available for model training, but performs much worse for out-of-distribution (OOD) samples. In real-world safety-critical applications, in particular, it is important to be aware if a new data point is OOD. To date, OOD detection is typically addressed using either confidence scores, auto-encoder based reconstruction, or contrastive learning. However, the global image context has not yet been explored to discriminate the non-local objectness between in-distribution and OOD samples. This paper proposes a first-of-its-kind OOD detection architecture named OODformer that leverages the contextualization capabilities of the transformer. Incorporating the transformer as the principal feature extractor allows us to exploit the object concepts and their discriminatory attributes along with their co-occurrence via visual attention. Based on contextualised embedding, we demonstrate OOD detection using both class-conditioned latent space similarity and a network confidence score. Our approach shows improved generalizability across various datasets. We have achieved a new state-of-the-art result on CIFAR-10/-100 and ImageNet30.
PublicationDomain Shifts in Reinforcement Learning: Identifying Disturbances in Environments( 2021)
; ;Schmoeller Roza, Felippe ; ;Günnemann, StephanEnd-to-End Deep Reinforcement Learning (RL) systems return an action no matter what situation they are confronted with, even for situations that differ entirely from those an agent has been trained for. In this work, we propose to formalize the changes in the environment in terms of the Markov Decision Process (MDP), resulting in a more formal framework when dealing with such problems.