Kaggle BirdCLEF+ 2026にてRank 34 / 4,092チーム（Top 0.8%・Silver Medal）を達成。音声データを用いた鳥類種識別コンペティション。
"}},{"node":{"frontmatter":{"title":"NVIDIA Nemotron Reasoning Challenge — Silver Medal","tech":["Python","LLM","LoRA Fine-tuning","PEFT"],"github":"","external":"https://www.kaggle.com/ngyzly"},"html":"NVIDIA Nemotron Model Reasoning ChallengeにてRank 65 / 4,182チーム（Top 1.5%・Silver Medal）を達成。LLMのLoRA効率的ファインチューニングによる推論能力強化。
"}},{"node":{"frontmatter":{"title":"Kaggle Santa 2025 — Silver Medal","tech":["Python","アルゴリズム最適化","組み合わせ最適化"],"github":"","external":"https://www.kaggle.com/ngyzly"},"html":"Kaggle Santa 2025 Christmas Tree Packing ChallengeにてRank 119 / 3,357チーム（Top 3.5%・Silver Medal）を達成。
"}},{"node":{"frontmatter":{"title":"SemiOrth — 直交デュアルネットワーク半教師あり学習","tech":["PyTorch","Semi-supervised Learning","MRI/CT Segmentation"],"github":"","external":"https://doi.org/10.1117/12.3047175"},"html":"強化された半教師あり医用画像セグメンテーションのための直交デュアルネットワークアーキテクチャSemiOrth。SPIE Medical Imaging 2025: Image Processing 採択。
"}},{"node":{"frontmatter":{"title":"Kaggle Deep Past — Akkadian翻訳 Silver Medal","tech":["Python","NLP","機械翻訳","Transformer"],"github":"","external":"https://www.kaggle.com/ngyzly"},"html":"Kaggle Deep Past Challenge（アッカド語→英語翻訳）にてRank 22 / 3,148チーム（Top 0.6%・Silver Medal）を達成。
"}},{"node":{"frontmatter":{"title":"Laparoscopic Video Segmentation","tech":["PyTorch","Contrastive Learning","Video Segmentation"],"github":"","external":"https://doi.org/10.1049/htl2.12069"},"html":"クラス単位対比学習とマルチスケール特徴抽出による腹腔鏡手術映像セグメンテーション手法。Healthcare Technology Letters 11.2-3 (2024) 採択。
"}},{"node":{"frontmatter":{"title":"Kaggle Vesuvius Challenge — Silver Medal","tech":["Python","PyTorch","CT画像解析","3D再構成"],"github":"","external":"https://www.kaggle.com/ngyzly"},"html":"Kaggle Vesuvius Challenge（ヘルクラネウム古代巻物解読）Surface DetectionにてRank 20 / 1,391チーム（Top 1.4%・Silver Medal）を達成。
"}},{"node":{"frontmatter":{"title":"MICCAI 2020 — COVID-19 CT Lesion Segmentation","tech":["Python","PyTorch","nnU-Net","CT画像解析","Semi-supervised Learning"],"github":"","external":"https://covid-segmentation.grand-challenge.org/"},"html":"MICCAI 2020 COVID-19 Lung CT Lesion Segmentation ChallengeにてRank **6 / 1,096チーム（Top 0.5%）**を達成。半教師あり学習と疑似ラベル手法を組み合わせ、少数アノテーションデータから高精度なCT肺病変領域の分割を実現。成果は国際学術誌 Medical Image Analysis（DOI: 10.1016/j.media.2022.102605）に掲載。
"}}]}}}