The document contains mathematical equations and notation related to machine learning and probability distributions. It involves defining terms like P(y|x), which represents the probability of outcome y given x, and exploring ways to calculate the expected value of an objective function Rn under different probability distributions p and q over the variables x and y. The goal appears to be to select parameters θ to optimize some objective while accounting for the distributions of the training data.
The document discusses the rights of data subjects under the EU GDPR, particularly regarding automated decision-making and profiling. It outlines conditions under which such decisions can be made, emphasizing the need for measures that protect the data subjects' rights and freedoms. Additionally, it includes references to various machine learning and artificial intelligence interpretability frameworks and studies.
Variational Template Machine for Data-to-Text Generationharmonylab
?
公開URL:https://openreview.net/forum?id=HkejNgBtPB
出典:Rong Ye, Wenxian Shi, Hao Zhou, Zhongyu Wei, Lei Li : Variational Template Machine for Data-to-Text Generation, 8th International Conference on Learning Representations(ICLR2020), Addis Ababa, Ethiopia (2020)
概要:Table形式の構造化データから文章を生成するタスク(Data-to-Text)において、Variational Auto Encoder(VAE)ベースの手法Variational Template Machine(VTM)を提案する論文です。Encoder-Decoderモデルを用いた既存のアプローチでは、生成文の多様性に欠けるという課題があります。本論文では多様な文章を生成するためにはテンプレートが重要であるという主張に基づき、テンプレートを学習可能なVAEベースの手法を提案します。提案手法では潜在変数の空間をテンプレート空間とコンテンツ空間に明示的に分離することによって、正確で多様な文生成が可能となります。また、table-textのペアデータだけではなくtableデータのないraw textデータを利用した半教師あり学習を行います。
Invariant Information Clustering for Unsupervised Image Classification and Se...harmonylab
?
紹介論文
Invariant Information Clustering for Unsupervised Image Classification and Segmentation
Xu J, Jo?o F. Henriques, Andrea Vedaldi
出典:Xu J, Jo?o F. Henriques, Andrea Vedaldi:Invariant Information Clustering forUnsupervised Image Classification and Segmentation, International Conference on Computer Vision (ICCV 2019), Seoul, Korea
概要:本論文では、正解ラベルを必要としない教師なし学習手法IICを提案しています。元画像に一般的なランダム変換を加えたペアを作成し、元画像とペアの相互情報量を最大化するよう学習を行います。画像のクラス分類?セグメンテーションタスクにおいて、8つのベンチマークでSOTAを達成しています。さらに、半教師あり学習にすることで、従来の教師あり学習精度を超える結果を得ています
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.
ArcFace: Additive Angular Margin Loss for Deep Face Recognitionharmonylab
?
出典: Jiankang Deng, Jia Guo, Niannan Xue, Stefanos Zafeiriou : ArcFace: Additive Angular Margin Loss for Deep Face Recognition, Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2019)
公開URL:https://arxiv.org/abs/1801.07698
概要 : 顔認識のための畳み込みニューラルネットワーク(DCNN)の課題は識別力を高める適切な損失関数を設計することです。本論文では、顔認識のための識別性の高い特徴量を得るために、Additive Angular Margin Loss (ArcFace)を提案します。一般的な顔認識ベンチマークから1兆ペアの大規模データセットなどを用いて、最先端顔認識技術との比較実験を行いました。結果は、従来手法を凌駕する精度を持つことが明らかになりました。
The document discusses the rights of data subjects under the EU GDPR, particularly regarding automated decision-making and profiling. It outlines conditions under which such decisions can be made, emphasizing the need for measures that protect the data subjects' rights and freedoms. Additionally, it includes references to various machine learning and artificial intelligence interpretability frameworks and studies.
Variational Template Machine for Data-to-Text Generationharmonylab
?
公開URL:https://openreview.net/forum?id=HkejNgBtPB
出典:Rong Ye, Wenxian Shi, Hao Zhou, Zhongyu Wei, Lei Li : Variational Template Machine for Data-to-Text Generation, 8th International Conference on Learning Representations(ICLR2020), Addis Ababa, Ethiopia (2020)
概要:Table形式の構造化データから文章を生成するタスク(Data-to-Text)において、Variational Auto Encoder(VAE)ベースの手法Variational Template Machine(VTM)を提案する論文です。Encoder-Decoderモデルを用いた既存のアプローチでは、生成文の多様性に欠けるという課題があります。本論文では多様な文章を生成するためにはテンプレートが重要であるという主張に基づき、テンプレートを学習可能なVAEベースの手法を提案します。提案手法では潜在変数の空間をテンプレート空間とコンテンツ空間に明示的に分離することによって、正確で多様な文生成が可能となります。また、table-textのペアデータだけではなくtableデータのないraw textデータを利用した半教師あり学習を行います。
Invariant Information Clustering for Unsupervised Image Classification and Se...harmonylab
?
紹介論文
Invariant Information Clustering for Unsupervised Image Classification and Segmentation
Xu J, Jo?o F. Henriques, Andrea Vedaldi
出典:Xu J, Jo?o F. Henriques, Andrea Vedaldi:Invariant Information Clustering forUnsupervised Image Classification and Segmentation, International Conference on Computer Vision (ICCV 2019), Seoul, Korea
概要:本論文では、正解ラベルを必要としない教師なし学習手法IICを提案しています。元画像に一般的なランダム変換を加えたペアを作成し、元画像とペアの相互情報量を最大化するよう学習を行います。画像のクラス分類?セグメンテーションタスクにおいて、8つのベンチマークでSOTAを達成しています。さらに、半教師あり学習にすることで、従来の教師あり学習精度を超える結果を得ています
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.
ArcFace: Additive Angular Margin Loss for Deep Face Recognitionharmonylab
?
出典: Jiankang Deng, Jia Guo, Niannan Xue, Stefanos Zafeiriou : ArcFace: Additive Angular Margin Loss for Deep Face Recognition, Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2019)
公開URL:https://arxiv.org/abs/1801.07698
概要 : 顔認識のための畳み込みニューラルネットワーク(DCNN)の課題は識別力を高める適切な損失関数を設計することです。本論文では、顔認識のための識別性の高い特徴量を得るために、Additive Angular Margin Loss (ArcFace)を提案します。一般的な顔認識ベンチマークから1兆ペアの大規模データセットなどを用いて、最先端顔認識技術との比較実験を行いました。結果は、従来手法を凌駕する精度を持つことが明らかになりました。
The document discusses various frequencies related to ball bearings, specifically ball spin frequency (bsf), fundamental train frequency (ftf), and ball passing frequency (bpfo). It mentions the application of Grad-CAM in conjunction with CNNs to analyze these frequencies, referencing multiple sources for related techniques. The context appears to involve signal processing and analysis in mechanical engineering.
Tandem connectionist anomaly detection: Use of faulty vibration signals in fe...pcl-lab
?
This document proposes a method called tandem connectionist anomaly detection that uses faulty vibration data from non-target machines to improve anomaly detection performance on a target machine. The method uses a deep neural network trained on both normal and faulty non-target data to learn discriminative features, which are then used as input to a Gaussian mixture model anomaly detector trained on normal target data. Experiments show this method significantly improves anomaly detection compared to using hand-crafted features or transferring just the detector. It demonstrates the ability to transfer the system between machines of both the same type and different types.
Bilinear map of filter-bank outputs for DNN-based speech recognitionpcl-lab
?
The document discusses the application of quadratic expansion (QE) of acoustic features in DNN-based speech recognition to improve phoneme error rates, as demonstrated through experimentation on the TIMIT dataset. The effectiveness of QE is highlighted, showing enhanced class separability and better representation of input features. Various DNN architectures and combinations of feature extraction methods are compared to establish the improvements in speech recognition performance.
Adaptive training of vibration-based anomaly detector for wind turbine condit...pcl-lab
?
The document discusses the development of an adaptive training method for an anomaly detector used in wind turbine condition monitoring, emphasizing the challenges of limited data availability. It presents a gmm-based anomaly detection system that can be effectively adapted using a small amount of data from the target device while leveraging a pre-trained model. The results indicate that this adaptive approach significantly improves anomaly detection performance, especially in early stages of monitoring.