Evaluations of large-scale recognition methods typically focus on overall performance. While this approach is common, it often fails to provide insights into performance across individual classes, which can lead to fairness issues and misrepresentation. Addressing these gaps is crucial for accurately assessing how well methods handle novel or unseen classes and ensuring a fair evaluation. To address fairness in Open-Set Recognition (OSR), we demonstrate that per-class performance can vary dramatically. We introduce Gaussian Hypothesis Open Set Technique (GHOST), a novel hyperparameter-free algorithm that models deep features using class-wise multivariate Gaussian distributions with diagonal covariance matrices. We apply Z-score normalization to logits to mitigate the impact of feature magnitudes that deviate from the model’s expectations, thereby reducing the likelihood of the network assigning a high score to an unknown sample. We evaluate GHOST across multiple ImageNet-1K pre-trained deep networks and test it with four different unknown datasets. Using standard metrics such as AUOSCR, AUROC and FPR95, we achieve statistically significant improvements, advancing the state-of-the-art in large-scale OSR. Source code is provided online.

Handling novelty remains a key challenge in visual recognition systems. Existing open-set recognition (OSR) methods rely on the familiarity hypothesis, detecting novelty by the absence of familiar features. We propose a novel attenuation hypothesis: small weights learned during training attenuate features and serve a dual role-differentiating known classes while discarding information useful for distinguishing known from unknown classes. To leverage this overlooked information, we present COSTARR, a novel approach that combines both the requirement of familiar features and the lack of unfamiliar ones. We provide a probabilistic interpretation of the COSTARR score, linking it to the likelihood of correct classification and belonging in a known class. To determine the individual contributions of the pre- and post-attenuated features to COSTARR's performance, we conduct ablation studies that show both pre-attenuated deep features and the underutilized post-attenuated Hadamard product features are essential for improving OSR. Also, we evaluate COSTARR in a large-scale setting using ImageNet2012-1K as known data and NINCO, iNaturalist, OpenImage-O, and other datasets as unknowns, across multiple modern pre-trained architectures (ViTs, ConvNeXts, and ResNet). The experiments demonstrate that COSTARR generalizes effectively across various architectures and significantly outperforms prior state-of-the-art methods by incorporating previously discarded attenuation information, advancing open-set recognition capabilities.

There exists a distribution discrepancy between training and testing, in the way images are fed to modern CNNs. Recent work tried to bridge this gap either by fine-tuning or re-training the network at different resolutions. However retraining a network is rarely cheap and not always viable. To this end, we propose a simple solution to address the train-test distributional shift and enhance the performance of pretrained models - which commonly ship as a package with deep learning platforms e.g., PyTorch. Specifically, we demonstrate that running inference on the center crop of an image is not always the best as important discriminatory information may be cropped-off. Instead we propose to combine results for multiple random crops for a test image. This not only matches the train time augmentation but also provides the full coverage of the input image. We explore combining representation of random crops through averaging at different levels i.e., deep feature level, logit level, and softmax level. We demonstrate that, for various families of modern deep networks, such averaging results in better validation accuracy compared to using a single central crop per image. The softmax averaging results in the best performance for various pre-trained networks without requiring any re-training or fine-tuning whatsoever. On modern GPUs with batch processing, the paper's approach to inference of pre-trained networks, is essentially free as all images in a batch can all be processed at once. Our code is available at: https://github.com/TouqeerAhmad/MID

Managing inputs that are novel, unknown, or out-of-distribution is critical as an agent moves from the lab to the open world. Novelty-related problems include being tolerant to novel perturbations of the normal input, detecting when the input includes novel items, and adapting to novel inputs. While significant research has been undertaken in these areas, a noticeable gap exists in the lack of a formalized definition of novelty that transcends problem domains. As a team of researchers spanning multiple research groups and different domains, we have seen, first hand, the difficulties that arise from ill-specified novelty problems, as well as inconsistent definitions and terminology. Therefore, we present the first unified framework for formal theories of novelty and use the framework to formally define a family of novelty types. Our framework can be applied across a wide range of domains, from symbolic AI to reinforcement learning, and beyond to open world image recognition. Thus, it can be used to help kick-start new research efforts and accelerate ongoing work on these important novelty-related problems.

We present Zillow Indoor Dataset (ZInD): A large indoor dataset with 71,474 panoramas from 1,524 real unfurnished homes. ZInD provides annotations of 3D room layouts, 2D and 3D floor plans, panorama location in the floor plan, and locations of windows and doors. The ground truth construction took over 1,500 hours of annotation work. To the best of our knowledge, ZInD is the largest real dataset with layout annotations. A unique property is the room layout data, which follows a real world distribution (cuboid, more general Manhattan, and non-Manhattan layouts) as opposed to the mostly cuboid or Manhattan layouts in current publicly available datasets. Also, the scale and annotations provided are valuable for effective research related to room layout and floor plan analysis. To demonstrate ZInD’s benefits, we benchmark on room layout estimation from single panoramas and multi-view registration.

Image classification in the open-world must handle out-of-distribution (OOD) images. Systems should ideally reject OOD images, or they will map atop of known classes and reduce reliability. Using open-set classifiers that can reject OOD inputs can help. However, optimal accuracy of open-set classifiers depend on the frequency of OOD data. Thus, for either standard or open-set classifiers, it is important to be able to determine when the world changes and increasing OOD inputs will result in reduced system reliability. However, during operations, we cannot directly assess accuracy as there are no labels. Thus, the reliability assessment of these classifiers must be done by human operators, made more complex because networks are not 100% accurate, so some failures are to be expected. To automate this process, herein, we formalize the open-world recognition reliability problem and propose multiple automatic reliability assessment policies to address this new problem using only the distribution of reported scores/probability data. The distributional algorithms can be applied to both classic classifiers with SoftMax as well as the open-world Extreme Value Machine (EVM) to provide automated reliability assessment. We show that all of the new algorithms significantly outperform detection using the mean of SoftMax.

In real-world deployments, machine learning applications find challenges when accessing ever-increasing volumes of data - the real world is open and often presents data from classes not seen in training. Open-set recognition is a growing area of machine learning addressing such problems. This research work advances the state-of-the-art in open-set recognition, the Extreme Value Machine (EVM), with a novel clustering-based extension (C-EVM) during training to improve the end-to-end prediction performance. The C-EVM combines Density-based spatial clustering of applications with noise (DBSCAN)-based clustering with a novel Nearby Clusters (NC) algorithm during model fitting to reduce computation while improving accuracy. Our experiments show a statistically significant improvement of 5-10% in macro F1-score over the state-of-the-art EVM on open-set testing using the KDD CUP-99 data set. Past work on open set recognition often traded improved open-set robustness for a decrease in closed-set accuracy, whereas C-EVM outperforms the EVM in both closed-set and open-set recognition. Testing on subsets of ImageNet-2012 with varying numbers of classes, the C-EVM statistically significantly out performs EVM when using deep features. A parameterless Hierarchical DBSCAN (HDBSCAN)-based C-EVM variant is introduced as part of this work that scales well for large data sets. Finally, both EVM and C-EVM can operate as kernel-free incremental learners, enabling these open-set multi-class classifiers to be useful for streaming and big data applications.

Tampered images pose a serious predicament since digitized media is a ubiquitous part of our lives. These are facilitated by the availability of image editing software and recent advances in deep Generative Adversarial Networks (GANs). We propose an innovative method to formulate the problem of 10-calizing manipulated regions in fake images as a deep representation learning problem using the Information Bottleneck (IB) principle. We devise a convolutional neural net-based architecture, InfoPrint (IP), that uses variational inference to approximate the IB formulation. Testing on three standard datasets, we demonstrate that InfoPrint outperforms the state-of-the-art by 3% points or more. Additionally, we demonstrate that it has the ability to to detect alterations made by inpainting GANs.

As science attempts to close the gap between man and machine by building systems capable of learning, we must embrace the importance of the unknown. The ability to differentiate between known and unknown can be considered a critical element of any intelligent self-learning system. The ability to reject uncertain inputs has a very long history in machine learning, as does including a background or garbage class to account for inputs that are not of interest. This paper explains why neither of these is genuinely sufficient for handling unknown inputs – uncertain is not unknown, and unknowns need not appear to be uncertain to a learning system. The past decade has seen the formalization and development of many open set algorithms, which provably bound the risk from unknown classes. We summarize the state of the art, core ideas, and results and explain why, despite the efforts to date, the current techniques are genuinely insufficient for handling unknown inputs, especially for deep networks.

Typically, most network intrusion detection systems use supervised learning techniques to identify network anomalies. A problem exists when identifying the unknowns and automatically updating a classifier with new query classes. This is defined as an open set incremental learning problem and we propose to extend a recently introduced method, the Extreme Value Machine (EVM) to address the issue of identifying new classes during query time. The EVM is derived from the statistical extreme value theory and is the first classifier that can perform kernel-free, nonlinear, variable bandwidth outlier detection combined with incremental learning. In this paper, we utilize the EVM for intrusion detection and measure the open set recognition performance of identifying known and unknown classes. Additionally, we evaluate the performance on the KDDCUP'99 dataset and compare the results with the state-of-the-art Weibull-SVM (W-SVM). Our findings demonstrate that the EVM mirrors the performance of the W-SVM classifier, while it supports incremental learning.