{"corpus_id":9136422,"paper_sha":"e76f13cc381a4ff9e10b1aa20f6750350fdd2632","doi":"10.1007/978-3-319-67534-3_7","arxiv_id":"1706.03509","pmid":null,"pmcid":null,"mag_id":2963983420,"dblp_id":"journals/corr/CheplyginaMVBP17","acl_id":null,"title":"Exploring the Similarity of Medical Imaging Classification Problems","year":2017,"publication_date":"2017-06-12","venue":"CVII-STENT/LABELS@MICCAI","journal":{"name":null,"pages":"59-66","volume":null},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle"],"pubmed_pub_types":null,"s2_fields_of_study":["Medicine","Computer Science","Mathematics"],"reference_count":15,"citation_count":13,"influential_citation_count":0,"is_open_access":false,"arxiv_categories":["cs.CV"],"arxiv_license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":null,"s2_open_access_landing_url":null,"s2_open_access_license":null,"s2_open_access_status":null,"pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"Supervised learning is ubiquitous in medical image analysis. In this paper we consider the problem of meta-learning – predicting which methods will perform well in an unseen classification problem, given previous experience with other classification problems. We investigate the first step of such an approach: how to quantify the similarity of different classification problems. We characterize datasets sampled from six classification problems by performance ranks of simple classifiers, and define the similarity by the inverse of Euclidean distance in this meta-feature space. We visualize the similarities in a 2D space, where meaningful clusters start to emerge, and show that the proposed representation can be used to classify datasets according to their origin with 89.3% accuracy. These findings, together with the observations of recent trends in machine learning, suggest that meta-learning could be a valuable tool for the medical imaging community.","claims":[{"public_id":"cl_ef870294b57fed7c0fac8baadb1f36c0","status":"active","text":"Classification-problem similarity can be quantified by the inverse of Euclidean distance in a meta-feature space built from performance ranks of simple classifiers.","confidence":0.97,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_ef870294b57fed7c0fac8baadb1f36c0"},{"public_id":"cl_0f7de53680ad1f624f6f9136b74b9100","status":"active","text":"Datasets can be classified by their origin with 89.3% accuracy using the proposed representation.","confidence":0.96,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_0f7de53680ad1f624f6f9136b74b9100"},{"public_id":"cl_80d48f5ea09170188d1f7beaa7fa8497","status":"active","text":"The findings support meta-learning as a potentially valuable tool for the medical imaging community.","confidence":0.83,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_80d48f5ea09170188d1f7beaa7fa8497"},{"public_id":"cl_dc7eaeb42c1f045742b1fcc84eeeea68","status":"active","text":"The proposed representation yields meaningful clusters in a two-dimensional visualization of problem similarities.","confidence":0.89,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_dc7eaeb42c1f045742b1fcc84eeeea68"}],"concepts":[{"public_id":"co_139b1bd3ba9d8059113a77eb061d218d","status":"active","name":"meta-learning","description":"A learning paradigm that uses experience from previous tasks to predict which methods will perform well on unseen tasks.","types":["method","learning paradigm"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_139b1bd3ba9d8059113a77eb061d218d"},{"public_id":"co_23bb947be5339f041782716f9918da06","status":"active","name":"meta-feature space","description":"A feature space constructed from summary characteristics of classification problems for comparing them at the problem level.","types":["representation","feature space"],"aliases":["meta-feature representation"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_23bb947be5339f041782716f9918da06"},{"public_id":"co_4140f94c35d0c874306dd337fc9f5368","status":"active","name":"performance ranks of simple classifiers","description":"Rank-based performance summaries of basic classifiers computed on datasets sampled from the classification problems.","types":["feature","summary statistic"],"aliases":["classifier performance ranks"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_4140f94c35d0c874306dd337fc9f5368"},{"public_id":"co_4452cf9f6c6300235b5a4c43292b9152","status":"active","name":"datasets","description":"The medical imaging datasets sampled from six classification problems and used in the analysis.","types":["dataset","data"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_4452cf9f6c6300235b5a4c43292b9152"},{"public_id":"co_52e23d984ed79eadce6e34997f134f63","status":"active","name":"89.3% accuracy","description":"The reported classification accuracy achieved when predicting dataset origin from the proposed representation.","types":["evaluation result","metric"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_52e23d984ed79eadce6e34997f134f63"},{"public_id":"co_77d53ae1f06e79f0771428b4ea143771","status":"active","name":"proposed representation","description":"The problem-level representation derived from classifier performance ranks and used for similarity analysis and origin classification.","types":["representation","method"],"aliases":["representation"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_77d53ae1f06e79f0771428b4ea143771"},{"public_id":"co_7a16554ca2b334f1ba5c12bb1fc91a12","status":"active","name":"meaningful clusters","description":"Groups that emerge in the similarity visualization and correspond to related classification problems.","types":["cluster","pattern"],"aliases":["clusters"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_7a16554ca2b334f1ba5c12bb1fc91a12"},{"public_id":"co_801e8437fc8859cbc2f678c40ee1cebf","status":"active","name":"Euclidean distance","description":"A geometric distance measure used here to compare points in the meta-feature space.","types":["distance measure","metric"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_801e8437fc8859cbc2f678c40ee1cebf"},{"public_id":"co_924b15b1c41defa5cc3400fb32dfcb9f","status":"active","name":"classification-problem similarity","description":"A measure of how alike two classification problems are in terms of their dataset behavior and model performance patterns.","types":["metric","concept"],"aliases":["similarity of classification problems"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_924b15b1c41defa5cc3400fb32dfcb9f"},{"public_id":"co_bae1f1c122d78c46a887c67f70ecbd5b","status":"active","name":"two-dimensional visualization","description":"A 2D embedding or plot used to display the relationships among classification problems.","types":["visualization","embedding"],"aliases":["2D space","2D visualization"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_bae1f1c122d78c46a887c67f70ecbd5b"},{"public_id":"co_bcff1bc0b62f7b5cf7a1e37a8e72f7a2","status":"active","name":"medical imaging community","description":"The research community concerned with developing and applying computational methods to medical images.","types":["community","field"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_bcff1bc0b62f7b5cf7a1e37a8e72f7a2"}],"external_ids":{"DOI":"10.1007/978-3-319-67534-3_7","ArXiv":"1706.03509","PubMed":null,"PubMedCentral":null,"MAG":2963983420,"DBLP":"journals/corr/CheplyginaMVBP17","ACL":null},"open_access":{"is_open_access":true,"pdf_url":"https://arxiv.org/pdf/1706.03509","landing_url":"https://arxiv.org/abs/1706.03509","source":"arxiv","pdf_url_source":"derived_arxiv","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","reason":null},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":true,"full_text_source":"arxiv","count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":631771,"paper_uid":"d6e7c78e-eef2-4ee3-a997-e87a632a7e3d","canonical_identity":{"paper_id":631771,"paper_uid":"d6e7c78e-eef2-4ee3-a997-e87a632a7e3d","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/9136422"}