[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-272852-105":59,"doc-detail-272852-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","semi-supervised-semantics-guided-adversarial-training-for-robust-trajectory-prediction","Semi-supervised Semantics-guided Adversarial Training for Robust Trajectory Prediction","","Trajectory prediction of surrounding objects is essential for self-driving and other autonomous systems, yet it remains vulnerable to adversarial attacks. Small, crafted perturbations to history trajectories can significantly mislead future predictions and trigger unsafe planning. This paper introduces SSAT, a semi-supervised semantics-guided adversarial training framework based on an adversarial autoencoder that learns disentangled semantic features and provides latent labels for adversarial supervision. Experiments across multiple attacks show robust mitigation up to 73% and improved generalization to unseen attack patterns, outperforming common defenses.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/semi-supervised-semantics-guided-adversarial-training-for-robust-trajectory-prediction/272852/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/semi-supervised-semantics-guided-adversarial-training-for-robust-trajectory-prediction/272852.png","ImageObject",300,407,{"name":92,"@type":93},"Mack Lee","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-27","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":46},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does the paper target in trajectory prediction?","Question",{"text":112,"@type":113},"It targets the adversarial robustness of vehicle trajectory prediction, where small crafted perturbations to history trajectories can cause unsafe future predictions and planning decisions.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does SSAT address the challenges of adversarial training for trajectory data?",{"text":117,"@type":113},"SSAT uses a semi-supervised adversarial autoencoder to model disentangled semantic features and generate additional latent labels, enabling effective adversarial training despite rich context and lack of class labels.",{"name":119,"@type":110,"acceptedAnswer":120},"What results are reported on defending against adversarial attacks?",{"text":121,"@type":113},"Experiments show SSAT can mitigate the impact of adversarial attacks by up to 73% and outperform other defense methods, while also improving robust generalization to unseen attack patterns.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},272852,1789468026,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":46,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":52,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":129,"read_time":143,"preview_page_count":4},1236954509970,"https://us-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","Semi-supervised Semantics-guided Adversarial Training for Robust Trajectory Prediction  \narXiv :2205 . 14230v2 [ cs .LG] 21 Mar 2023  \nRuochen Jiao Northwestern University  \n[ruochen.jiao@u.northwestern.edu](ruochen.jiao@u.northwestern.edu)  \nXiangguo Liu Northwestern University  \n[xg.liu@u.northwestern.edu](xg.liu@u.northwestern.edu)  \nTakami Sato University of California, Irvine  \n[takamis@uci.edu](takamis@uci.edu)  \nQi Alfred Chen University of California, Irvine  \n[alfchen@uci.edu](alfchen@uci.edu)  \nQi Zhu Northwestern University [qzhu@northwestern.edu](qzhu@northwestern.edu)  \nAbstract  \nPredicting the trajectories of surrounding objects is a critical task for self-driving vehicles and many other autonomous systems. Recent works demonstrate that adversarial attacks on trajectory prediction, where small crafted perturbations are introduced to history trajectories, may significantly mislead the prediction of future trajectories and induce unsafe planning. However, few works have addressed enhancing the robustness of this important safetycritical task. In this paper, we present a novel adversarial training method for trajectory prediction. Compared with typical adversarial training on image tasks, our work is challenged by more random input with rich context anda lack of class labels. To address these challenges, we propose a method based on a semi-supervised adversarial autoencoder, which models disentangled semantic features with domain knowledge and provides additional latent labels for the adversarial training. Extensive experiments with different types of attacks demonstrate that our Semisupervised Semantics-guided Adversarial Training (SSAT) method can effectively mitigate the impact of adversarial attacks by up to 73% and outperform other popular defense methods. In addition, experiments show that our method can significantly improve the system's robust generalization to unseen patterns of attacks. We believe that such semantics-guided architecture and advancement on robust generalization is an important step for developing robust prediction models and enabling safe decision making.  \n1. Introduction  \nThe adversarial robustness of deep neural networks (DNNs) has drawn significant attention in recent years. State-of-the-art classifiers can be misled to make erro-  \nneous predictions by slight but carefully-optimized perturbations [30] . Besides classification, adversarial attacks are also targeted at various other tasks such as object detection, image generation and natural language processing. To defend against adversarial examples, adversarial training is commonly used to enhance the intrinsic robustness of models and has been shown to be very effective among various defense strategies [16, 17, 26] . On the other hand, the works in [22, 33, 28] show that adversarial training may suffer poor robust generalization on unseen attacks and [25, 16, 29] demonstrate a trade-off between robustness and accuracy. In this paper, we address trajectory prediction, a safety-critical task that is common in self-driving vehicles and other autonomous systems, and propose a novel architecture to enhance its adversarial robustness and robust generalization by introducing semantic features and a semi-supervised adversarial autoencoder (AAE) [18] into the adversarial training process.  \nIn particular, we focus on the trajectory prediction module in autonomous driving, where large DNNs have enabled breakthroughs in recent years. More specifically, an autonomous driving system typically consists of several modules such as perception, localization, prediction, planning (route, behavior, trajectory, and motion planning), and control. The perception module is to detect agents and obstacles based on the sensing inputs (e.g., images and 3D point clouds), and the effectiveness and robustness of the perception module are well-studied in the research community. The prediction module is to predict future trajectories of surrounding vehicles based on the","cbCaidy6KWgnsRYz","https://ap.wps.com/l/cbCaidy6KWgnsRYz","pdf",941655,"English","# Introduction\n## Adversarial robustness and defense approaches\n## Trajectory prediction in autonomous driving\n## Threat model of adversarial trajectories","[{\"question\":\"What problem does the paper target in trajectory prediction?\",\"answer\":\"It targets the adversarial robustness of vehicle trajectory prediction, where small crafted perturbations to history trajectories can cause unsafe future predictions and planning decisions.\"},{\"question\":\"How does SSAT address the challenges of adversarial training for trajectory data?\",\"answer\":\"SSAT uses a semi-supervised adversarial autoencoder to model disentangled semantic features and generate additional latent labels, enabling effective adversarial training despite rich context and lack of class labels.\"},{\"question\":\"What results are reported on defending against adversarial attacks?\",\"answer\":\"Experiments show SSAT can mitigate the impact of adversarial attacks by up to 73% and outperform other defense methods, while also improving robust generalization to unseen attack patterns.\"}]","Semi-supervised Semantics-guided Adversarial Training for Robust Trajectory Prediction | PDF",25]