Semi supervised, weakly-supervised, unsupervised, and active learningYusuke Uchida
An overview of semi supervised learning, weakly-supervised learning, unsupervised learning, and active learning.
Focused on recent deep learning-based image recognition approaches.
Semi supervised, weakly-supervised, unsupervised, and active learningYusuke Uchida
An overview of semi supervised learning, weakly-supervised learning, unsupervised learning, and active learning.
Focused on recent deep learning-based image recognition approaches.
You Only Look One-level Featureの解説と見せかけた物体検出のよもやま話Yusuke Uchida
第7回全日本コンピュータビジョン勉強会「CVPR2021読み会」(前編)の発表資料です
https://kantocv.connpass.com/event/216701/
You Only Look One-level Featureの解説と、YOLO系の雑談や、物体検出における関連する手法等を広く説明しています
1. The document discusses research on time-series big data feature extraction and real-time forecasting.
2. The research aims to predict the future by analyzing large-scale data in order to transform society through optimizing social activities in real-time.
3. Key areas of focus include tensor analysis of complex time-stamped event data, non-linear modeling of non-linear social phenomena in big data, and real-time processing.
This document discusses some of the challenges in developing AI systems that utilize machine learning. It notes that machine learning systems rely on probabilities and statistics based on training data, making quality assurance difficult. It is also difficult to fully understand and interpret models from deep neural networks. The document suggests that new approaches are needed for developing machine learning-based systems, as traditional software engineering approaches do not work well. Establishing the field of "machine learning engineering" is important for building AI systems that can reliably ensure quality.
Several recent papers have explored self-supervised learning methods for vision transformers (ViT). Key approaches include:
1. Masked prediction tasks that predict masked patches of the input image.
2. Contrastive learning using techniques like MoCo to learn representations by contrasting augmented views of the same image.
3. Self-distillation methods like DINO that distill a teacher ViT into a student ViT using different views of the same image.
4. Hybrid approaches that combine masked prediction with self-distillation, such as iBOT.
You Only Look One-level Featureの解説と見せかけた物体検出のよもやま話Yusuke Uchida
第7回全日本コンピュータビジョン勉強会「CVPR2021読み会」(前編)の発表資料です
https://kantocv.connpass.com/event/216701/
You Only Look One-level Featureの解説と、YOLO系の雑談や、物体検出における関連する手法等を広く説明しています
1. The document discusses research on time-series big data feature extraction and real-time forecasting.
2. The research aims to predict the future by analyzing large-scale data in order to transform society through optimizing social activities in real-time.
3. Key areas of focus include tensor analysis of complex time-stamped event data, non-linear modeling of non-linear social phenomena in big data, and real-time processing.
This document discusses some of the challenges in developing AI systems that utilize machine learning. It notes that machine learning systems rely on probabilities and statistics based on training data, making quality assurance difficult. It is also difficult to fully understand and interpret models from deep neural networks. The document suggests that new approaches are needed for developing machine learning-based systems, as traditional software engineering approaches do not work well. Establishing the field of "machine learning engineering" is important for building AI systems that can reliably ensure quality.
Several recent papers have explored self-supervised learning methods for vision transformers (ViT). Key approaches include:
1. Masked prediction tasks that predict masked patches of the input image.
2. Contrastive learning using techniques like MoCo to learn representations by contrasting augmented views of the same image.
3. Self-distillation methods like DINO that distill a teacher ViT into a student ViT using different views of the same image.
4. Hybrid approaches that combine masked prediction with self-distillation, such as iBOT.
オープンコミュニティ「要求開発アライアンス」(http://www.openthology.org)の2011年6月定例会発表資料です。
Open Community "Requirement Development Alliance" 2011/6 regular meeting of the presentation materials.
2019年07月09日 リカレントエデュケーション講座@京橋。
楽天ではどのようにビッグデータを活用しているのか、データサイエンス&AIの最新応用事例の紹介。
およびデータサイエンス系のプロジェクトの進め方と,必要な役割についての紹介。
登壇者:平手勇宇(Rakuten Institute of Technology Tokyo)
201024 ai koeln (akemi yokota) auf japanischAkemi Yokota
This Slide is for the Symposium "TECHNICAL AND ETHICAL ASPECTS OF ARTIFICIAL INTELLIGENCE IN JAPAN AND GERMANY". Here is the original Version in Japanese. In the Symposium I will use a german version with the help of JKI center in Koeln.
ケルン日本文化会館シンポジウム「日独両国におけるデジタル化の諸相」での報告「AI利活用社会のための法制度設計 ~日本の状況と未来の展望」スライドの日本語版です。当日会場ではドイツ語版が投影されます(横田は日本語で話し,同時通訳あり)ので、日本語話者はこちらを参照してください。
19. AI プロダクト品質保証ガイドライン
CDLE LT2 @NHigashino
19
1 目的とスコープ
1.1 背景と目的
1.2 AI プロダクトの品質保証上の課題と本ガイドラインのスコープ
2 AI プロダクトの品質保証の枠組み
2.1 AI プロダクトの品質保証の基本的考え方
2.1.1 AI プロダクトの品質保証において考慮すべき軸
2.1.2 Data Integrity
2.1.3 Model Robustness
2.1.4 System Quality
2.1.5 Process Agility
2.1.6 Customer Expectation
2.2 AI プロダクトの品質保証の分類軸ごとのチェックリスト
2.2.1 Data Integrity
2.2.2 Model Robustness
2.2.3 System Quality
2.2.4 Process Agility
2.2.5 Customer Expectation
2.3 AI プロダクトの品質保証の構築・評価
2.3.1 バランスに着目した構築・評価
2.3.2 開発段階に着目した構築・評価
2.3.3 余力と過剰品質
3 技術カタログ
3.1 AI プロダクト固有の品質特性
3.1.1 教師あり学習のモデルに対する性能指標
3.1.2 データに対する評価
3.1.3 頑健性
3.1.4 公平性
2020/9/30