More Like this

1. InvertibleNetworks.jl: A Julia package for scalablenormalizing flows

   https://doi.org/10.21105/joss.06554  

   Orozco, Rafael; Witte, Philipp; Louboutin, Mathias; Siahkoohi, Ali; Rizzuti, Gabrio; Peters, Bas; Herrmann, Felix J (July 2024, Journal of Open Source Software)

   Normalizing flows is a density estimation method that provides efficient exact likelihood estimation and sampling (Dinh et al., 2014) from high-dimensional distributions. This method depends on the use of the change of variables formula, which requires an invertible transform. Thus normalizing flow architectures are built to be invertible by design (Dinh et al., 2014). In theory, the invertibility of architectures constrains the expressiveness, but the use of coupling layers allows normalizing flows to exploit the power of arbitrary neural networks, which do not need to be invertible, (Dinh et al., 2016) and layer invertibility means that, if properly implemented, many layers can be stacked to increase expressiveness without creating a training memory bottleneck. The package we present, InvertibleNetworks.jl, is a pure Julia (Bezanson et al., 2017) imple- mentation of normalizing flows. We have implemented many relevant neural network layers, including GLOW 1x1 invertible convolutions (Kingma & Dhariwal, 2018), affine/additive coupling layers (Dinh et al., 2014), Haar wavelet multiscale transforms (Haar, 1909), and Hierarchical invertible neural transport (HINT) (Kruse et al., 2021), among others. These modular layers can be easily composed and modified to create different types of normalizing flows. As starting points, we have implemented RealNVP, GLOW, HINT, Hyperbolic networks (Lensink et al., 2022) and their conditional counterparts for users to quickly implement their individual applications. 

   more » « less
2. Improving Out-of-Distribution Detection using Segmented Images and Cross-View Attention Fusion

   Politowicz, A; Mazumder, S; Liu, B (March 2026, Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision)

   Although out-of-distribution (OOD) detection has been extensively studied, it continues to face challenges in handling OOD data semantically similar to In-Distribution (ID) data. Part of the difficulty arises from the model's inability to learn superior ID class discriminative features. We propose to improve this by segmenting input images into foreground and background views and combining them with the original input image (original view) in a multi-view learning approach. We present a novel method, called CASOD (Cross-view Attention of Segmented views for OOD Detection), that learns better discriminative information from all three views and subsequently, fuses them through a novel stacked cross-view attention mechanism to produce the final predictive feature representation. A feature-based method is then applied to the final fused feature for OOD detection, giving major improvements over a range of strong baselines on various near-and far-OOD datasets. CASOD achieves state-of-the-art performance in various experimental settings with challenging ID and OOD datasets. Code for CASOD can be found at https://github. com/alexp205/CASOD. 

   more » « less
3. Robustness to Spurious Correlations Improves Semantic Out-of-Distribution Detection

   https://doi.org/10.1609/aaai.v37i12.26785  

   Zhang, Lily H.; Ranganath, Rajesh (January 2023, AAAI, 2023.)

   Methods which utilize the outputs or feature representations of predictive models have emerged as promising approaches for out-of-distribution (OOD) detection of image inputs. However, these methods struggle to detect OOD inputs that share nuisance values (e.g. background) with in-distribution inputs. The detection of shared-nuisance out-of-distribution (SN-OOD) inputs is particularly relevant in real-world applications, as anomalies and in-distribution inputs tend to be captured in the same settings during deployment. In this work, we provide a possible explanation for SN-OOD detection failures and propose nuisance-aware OOD detection to address them. Nuisance-aware OOD detection substitutes a classifier trained via empirical risk minimization and cross-entropy loss with one that 1. is trained under a distribution where the nuisance-label relationship is broken and 2. yields representations that are independent of the nuisance under this distribution, both marginally and conditioned on the label. We can train a classifier to achieve these objectives using Nuisance-Randomized Distillation (NuRD), an algorithm developed for OOD generalization under spurious correlations. Output- and feature-based nuisance-aware OOD detection perform substantially better than their original counterparts, succeeding even when detection based on domain generalization algorithms fails to improve performance. 

   more » « less
4. Training OOD Detectors in their Natural Habitats

   Katz-Samuels, Julian; Nakhleh, Julia; Nowak, Robert; Li, Yixuan (January 2022, International Conference on Machine Learning)

   Out-of-distribution (OOD) detection is important for machine learning models deployed in the wild. Recent methods use auxiliary outlier data to regularize the model for improved OOD detection. However, these approaches make a strong distributional assumption that the auxiliary outlier data is completely separable from the in-distribution (ID) data. In this paper, we propose a novel framework that leverages wild mixture data -- that naturally consists of both ID and OOD samples. Such wild data is abundant and arises freely upon deploying a machine learning classifier in their \emph{natural habitats}. Our key idea is to formulate a constrained optimization problem and to show how to tractably solve it. Our learning objective maximizes the OOD detection rate, subject to constraints on the classification error of ID data and on the OOD error rate of ID examples. We extensively evaluate our approach on common OOD detection tasks and demonstrate superior performance. 

   more » « less
5. Turning the Curse of Heterogeneity in Federated Learning into a Blessing for Out-of-Distribution Detection

   Yu, Shuyang; Hong, Junyuan; Wang, Haotao; Wang, Zhangyang; Zhou, Jiayu (April 2023, International Conference on Learning Representations (ICLR))

   Deep neural networks have witnessed huge successes in many challenging prediction tasks and yet they often suffer from out-of-distribution (OoD) samples, misclassifying them with high confidence. Recent advances show promising OoD detection performance for centralized training, and however, OoD detection in federated learning (FL) is largely overlooked, even though many security sensitive applications such as autonomous driving and voice recognition authorization are commonly trained using FL for data privacy concerns. The main challenge that prevents previous state-of-the-art OoD detection methods from being incorporated to FL is that they require large amount of real OoD samples. However, in real-world scenarios, such large-scale OoD training data can be costly or even infeasible to obtain, especially for resource-limited local devices. On the other hand, a notorious challenge in FL is data heterogeneity where each client collects non-identically and independently distributed (non-iid) data. We propose to take advantage of such heterogeneity and turn the curse into a blessing that facilitates OoD detection in FL. The key is that for each client, non-iid data from other clients (unseen external classes) can serve as an alternative to real OoD samples. Specifically, we propose a novel Federated Out-of-Distribution Synthesizer (FOSTER), which learns a class-conditional generator to synthesize virtual external-class OoD samples, and maintains data confidentiality and communication efficiency required by FL. Experimental results show that our method outperforms the state-of-the-art by 2.49%, 2.88%, 1.42% AUROC, and 0.01%, 0.89%, 1.74% ID accuracy, on CIFAR-10, CIFAR-100, and STL10, respectively. 

   more » « less