3D-ICE-arXiv / Papers

3 stars 1 forks source link

Difficulty-Aware Meta-learning for Rare Disease Diagnosis #51

Open 3D-ICE-arXiv opened 3 years ago

3D-ICE-arXiv commented 3 years ago

Department of Computer Science and Engineering, The Chinese University of Hong Kong Department of Radiation Oncology, Stanford University Shenzhen Key Laboratory of Virtual Reality and Human Interaction Technology

■どんなもの? 希少疾患を対象として、データが少なくても何とか対処しようとする試みについての論文。 difficulty-aware meta-learning methodを提案。

■先行研究と比べてどこがすごい?

■技術や手法のキモはどこ? 最初にメタ学習を訓練 一般的な疾患のデータからモデルを作成し、モデルを適応させて希少疾患の分類に対応 難易度を考慮した学習の重要性を動的に監視するメタ学習法タスクを作成することで以上を成し遂げている。

■どうやって有効だと検証した? 最近のISIC 2018皮膚病変分類データセットを使用 AUC of 83.3%を示した。

■議論はある?

■次に読むべき論文は?

image

image

3D-ICE-arXiv commented 3 years ago

https://github.com/3D-ICE-arXiv/Papers/issues/12