Semantic Novelty Detection via Relational Reasoning. Borlino, F., C., Bucci, S., & Tommasi, T. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 13685 LNCS:183-200, 2022. ![Semantic Novelty Detection via Relational Reasoning [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Paper doi abstract bibtex Semantic novelty detection aims at discovering unknown categories in the test data. This task is particularly relevant in safety-critical applications, such as autonomous driving or healthcare, where it is crucial to recognize unknown objects at deployment time and issue a warning to the user accordingly. Despite the impressive advancements of deep learning research, existing models still need a finetuning stage on the known categories in order to recognize the unknown ones. This could be prohibitive when privacy rules limit data access, or in case of strict memory and computational constraints (e.g. edge computing). We claim that a tailored representation learning strategy may be the right solution for effective and efficient semantic novelty detection. Besides extensively testing state-of-the-art approaches for this task, we propose a novel representation learning paradigm based on relational reasoning. It focuses on learning how to measure semantic similarity rather than recognizing known categories. Our experiments show that this knowledge is directly transferable to a wide range of scenarios, and it can be exploited as a plug-and-play module to convert closed-set recognition models into reliable open-set ones.
@article{
title = {Semantic Novelty Detection via Relational Reasoning},
type = {article},
year = {2022},
keywords = {Domain generalization,Novelty detection,Open set learning,Relational reasoning,Representation learning},
pages = {183-200},
volume = {13685 LNCS},
id = {b833397b-81a2-3494-9ed8-83832f4cc92c},
created = {2023-04-21T09:42:40.792Z},
file_attached = {true},
profile_id = {bfbbf840-4c42-3914-a463-19024f50b30c},
group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},
last_modified = {2023-04-21T09:43:07.755Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
folder_uuids = {4b11b59b-846d-45b5-82b7-81d8bc3108b2},
private_publication = {false},
abstract = {Semantic novelty detection aims at discovering unknown categories in the test data. This task is particularly relevant in safety-critical applications, such as autonomous driving or healthcare, where it is crucial to recognize unknown objects at deployment time and issue a warning to the user accordingly. Despite the impressive advancements of deep learning research, existing models still need a finetuning stage on the known categories in order to recognize the unknown ones. This could be prohibitive when privacy rules limit data access, or in case of strict memory and computational constraints (e.g. edge computing). We claim that a tailored representation learning strategy may be the right solution for effective and efficient semantic novelty detection. Besides extensively testing state-of-the-art approaches for this task, we propose a novel representation learning paradigm based on relational reasoning. It focuses on learning how to measure semantic similarity rather than recognizing known categories. Our experiments show that this knowledge is directly transferable to a wide range of scenarios, and it can be exploited as a plug-and-play module to convert closed-set recognition models into reliable open-set ones.},
bibtype = {article},
author = {Borlino, Francesco Cappio and Bucci, Silvia and Tommasi, Tatiana},
doi = {10.1007/978-3-031-19806-9_11},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}
}
Downloads: 0
{"_id":"hFAqu7mkcsQnfbWtw","bibbaseid":"borlino-bucci-tommasi-semanticnoveltydetectionviarelationalreasoning-2022","author_short":["Borlino, F., C.","Bucci, S.","Tommasi, T."],"bibdata":{"title":"Semantic Novelty Detection via Relational Reasoning","type":"article","year":"2022","keywords":"Domain generalization,Novelty detection,Open set learning,Relational reasoning,Representation learning","pages":"183-200","volume":"13685 LNCS","id":"b833397b-81a2-3494-9ed8-83832f4cc92c","created":"2023-04-21T09:42:40.792Z","file_attached":"true","profile_id":"bfbbf840-4c42-3914-a463-19024f50b30c","group_id":"5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1","last_modified":"2023-04-21T09:43:07.755Z","read":false,"starred":false,"authored":false,"confirmed":"true","hidden":false,"folder_uuids":"4b11b59b-846d-45b5-82b7-81d8bc3108b2","private_publication":false,"abstract":"Semantic novelty detection aims at discovering unknown categories in the test data. This task is particularly relevant in safety-critical applications, such as autonomous driving or healthcare, where it is crucial to recognize unknown objects at deployment time and issue a warning to the user accordingly. Despite the impressive advancements of deep learning research, existing models still need a finetuning stage on the known categories in order to recognize the unknown ones. This could be prohibitive when privacy rules limit data access, or in case of strict memory and computational constraints (e.g. edge computing). We claim that a tailored representation learning strategy may be the right solution for effective and efficient semantic novelty detection. Besides extensively testing state-of-the-art approaches for this task, we propose a novel representation learning paradigm based on relational reasoning. It focuses on learning how to measure semantic similarity rather than recognizing known categories. Our experiments show that this knowledge is directly transferable to a wide range of scenarios, and it can be exploited as a plug-and-play module to convert closed-set recognition models into reliable open-set ones.","bibtype":"article","author":"Borlino, Francesco Cappio and Bucci, Silvia and Tommasi, Tatiana","doi":"10.1007/978-3-031-19806-9_11","journal":"Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)","bibtex":"@article{\n title = {Semantic Novelty Detection via Relational Reasoning},\n type = {article},\n year = {2022},\n keywords = {Domain generalization,Novelty detection,Open set learning,Relational reasoning,Representation learning},\n pages = {183-200},\n volume = {13685 LNCS},\n id = {b833397b-81a2-3494-9ed8-83832f4cc92c},\n created = {2023-04-21T09:42:40.792Z},\n file_attached = {true},\n profile_id = {bfbbf840-4c42-3914-a463-19024f50b30c},\n group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},\n last_modified = {2023-04-21T09:43:07.755Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {true},\n hidden = {false},\n folder_uuids = {4b11b59b-846d-45b5-82b7-81d8bc3108b2},\n private_publication = {false},\n abstract = {Semantic novelty detection aims at discovering unknown categories in the test data. This task is particularly relevant in safety-critical applications, such as autonomous driving or healthcare, where it is crucial to recognize unknown objects at deployment time and issue a warning to the user accordingly. Despite the impressive advancements of deep learning research, existing models still need a finetuning stage on the known categories in order to recognize the unknown ones. This could be prohibitive when privacy rules limit data access, or in case of strict memory and computational constraints (e.g. edge computing). We claim that a tailored representation learning strategy may be the right solution for effective and efficient semantic novelty detection. Besides extensively testing state-of-the-art approaches for this task, we propose a novel representation learning paradigm based on relational reasoning. It focuses on learning how to measure semantic similarity rather than recognizing known categories. Our experiments show that this knowledge is directly transferable to a wide range of scenarios, and it can be exploited as a plug-and-play module to convert closed-set recognition models into reliable open-set ones.},\n bibtype = {article},\n author = {Borlino, Francesco Cappio and Bucci, Silvia and Tommasi, Tatiana},\n doi = {10.1007/978-3-031-19806-9_11},\n journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}\n}","author_short":["Borlino, F., C.","Bucci, S.","Tommasi, T."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/19c54d58-f51e-212b-7374-addd9fd3a65a/220708699v2.pdf.pdf"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"borlino-bucci-tommasi-semanticnoveltydetectionviarelationalreasoning-2022","role":"author","keyword":["Domain generalization","Novelty detection","Open set learning","Relational reasoning","Representation learning"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"article","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["2252seNhipfTmjEBQ"],"keywords":["domain generalization","novelty detection","open set learning","relational reasoning","representation learning"],"search_terms":["semantic","novelty","detection","via","relational","reasoning","borlino","bucci","tommasi"],"title":"Semantic Novelty Detection via Relational Reasoning","year":2022}