Obtaining measured Synthetic Aperture Radar (SAR) data for training Automatic Target Recognition (ATR) models can be too expensive (in terms of time and money) and complex of a process in many situations. In response, researchers have developed methods for creating synthetic SAR data for targets using electromagnetic prediction software, which is then used to enrich an existing measured training dataset. However, this approach relies on the availability of some amount of measured data. In this work, we focus on the case of having 100% synthetic training data, while testing on only measured data. We use the SAMPLE dataset public released by AFRL, and find significant challenges to learning generalizable representations from the synthetic data due to distributional differences between the two modalities and extremely limited training sample quantities. Using deep learning-based ATR models, we propose data augmentation, model construction, loss function choices, and ensembling techniques to enhance the representation learned from the synthetic data, and ultimately achieved over 95% accuracy on the SAMPLE dataset. We then analyze the functionality of our ATR models using saliency and feature-space investigations and find them to learn a more cohesive representation of the measured and synthetic data. Finally, we evaluate the out-oflibrary detection performance of our synthetic-only models and find that they are nearly 10% more effective than baseline methods at identifying measured test samples that do not belong to the training class set. Overall, our techniques and their compositions significantly enhance the feasibility of using ATR models trained exclusively on synthetic data.
Training deep learning-based Synthetic Aperture Radar Automatic Target Recognition (SAR-ATR) systems for use in an "open-world" operating environment has thus far proven difficult. Most SAR-ATR systems are designed to achieve maximum accuracy for a limited set of classes, yet ignore the implications of encountering novel target classes during deployment. Even worse, the standard deep learning training objectives fundamentally inherit a closed-world assumption, and provide no guidance for how to handle out-of-distribution (OOD) data. In this work, we develop a novel training procedure called Adversarial Outlier Exposure (AdvOE) to co-design the ATR system for accuracy and OOD detection. Our method introduces a large, diverse and unlabeled auxiliary training dataset containing samples from the OOD set. The AdvOE objective encourages a Deep Neural Network to learn robust features of the in-distribution training data, while also promoting maximum entropy predictions for adversarially perturbed versions of the OOD data. We experiment with the recent SAMPLE dataset, and find our method nearly doubles OOD detection performance over the baseline in key settings, and excels when using only synthetic training data. As compared to several other advanced ATR training techniques, AdvOE also affords significant improvements in both classification and detection statistics. Finally, we conduct extensive experiments that measure the effect of OOD set granularity on detection rates; discuss the implications of using different detection algorithms; and develop a novel analysis technique to validate our findings and interpret the OOD detection problem from a new perspective.
The success of deep learning research has catapulted deep models into production systems that our society is becoming increasingly dependent on, especially in the image and video domains. However, recent work has shown that these largely uninterpretable models exhibit glaring security vulnerabilities in the presence of an adversary. In this work, we develop a powerful untargeted adversarial attack for action recognition systems in both white-box and black-box settings. Action recognition models differ from image-classification models in that their inputs contain a temporal dimension, which we explicitly target in the attack. Drawing inspiration from image classifier attacks, we create new attacks which achieve state-of-the-art success rates on a two-stream classifier trained on the UCF-101 dataset. We find that our attacks can significantly degrade a model's performance with sparsely and imperceptibly perturbed examples. We also demonstrate the transferability of our attacks to black-box action recognition systems.
Over recent years, a myriad of novel convolutional network architectures have been developed to advance stateof-the-art performance on challenging recognition tasks. As computational resources improve, a great deal of effort has been placed in efficiently scaling up existing designs and generating new architectures with Neural Architecture Search (NAS) algorithms. While network topology has proven to be a critical factor for model performance, we show that significant gains are being left on the table by keeping topology static at inference-time. Due to challenges such as scale variation, we should not expect static models configured to perform well across a training dataset to be optimally configured to handle all test data. In this work, we seek to expose the exciting potential of inference-timedynamic models. By allowing just four layers to dynamically change configuration at inference-time, we show that existing off-the-shelf models like ResNet-50 are capable of over 95% accuracy on ImageNet. This level of performance currently exceeds that of models with over 20x more parameters and significantly more complex training procedures.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.