機械学習の社会実装では、予測精度が高くても、機械学習がブラックボックであるために使うことができないということがよく起きます。
このスライドでは機械学習が不得意な予測結果の根拠を示すために考案されたLIMEの論文を解説します。
Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin. "" Why should i trust you?" Explaining the predictions of any classifier." Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. 2016.
1. The document introduces an example of applying incremental learning to anomaly detection in manufacturing processes. Incremental learning allows machine learning models to automatically adapt to environmental changes.
2. A case study is described where incremental learning was used for video-based anomaly detection on a work conveying device to detect abnormalities and stop equipment. This allowed the system to better adapt to state changes with fewer false alarms.
3. The benefits of incremental learning for adapting to changes over time without requiring retraining on new data are summarized. Potential issues like increased computation and inability to detect certain types of gradual changes are also noted.
This document contains a summary of 3 papers on deep residual networks and squeeze-and-excitation networks:
1. Kaiming He et al. "Deep Residual Learning for Image Recognition" which introduced residual networks for image recognition.
2. Andreas Veit et al. "Residual Networks Behave Like Ensembles of Relatively Shallow Networks" which analyzed how residual networks behave like ensembles.
3. Jie Hu et al. "Squeeze-and-Excitation Networks" which introduced squeeze-and-excitation blocks to help convolutional networks learn channel dependencies.
The document also references the PyTorch ResNet implementation and provides URLs to the first and third papers. It contains non-English
機械学習の社会実装では、予測精度が高くても、機械学習がブラックボックであるために使うことができないということがよく起きます。
このスライドでは機械学習が不得意な予測結果の根拠を示すために考案されたLIMEの論文を解説します。
Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin. "" Why should i trust you?" Explaining the predictions of any classifier." Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. 2016.
1. The document introduces an example of applying incremental learning to anomaly detection in manufacturing processes. Incremental learning allows machine learning models to automatically adapt to environmental changes.
2. A case study is described where incremental learning was used for video-based anomaly detection on a work conveying device to detect abnormalities and stop equipment. This allowed the system to better adapt to state changes with fewer false alarms.
3. The benefits of incremental learning for adapting to changes over time without requiring retraining on new data are summarized. Potential issues like increased computation and inability to detect certain types of gradual changes are also noted.
This document contains a summary of 3 papers on deep residual networks and squeeze-and-excitation networks:
1. Kaiming He et al. "Deep Residual Learning for Image Recognition" which introduced residual networks for image recognition.
2. Andreas Veit et al. "Residual Networks Behave Like Ensembles of Relatively Shallow Networks" which analyzed how residual networks behave like ensembles.
3. Jie Hu et al. "Squeeze-and-Excitation Networks" which introduced squeeze-and-excitation blocks to help convolutional networks learn channel dependencies.
The document also references the PyTorch ResNet implementation and provides URLs to the first and third papers. It contains non-English
This document appears to be a technical manual or guide from Fuji Xerox covering various topics related to printer functions and operations. It includes copyright information on the first page and contains numbered sections with formatting and images that suggest it is explaining settings and processes. The document provides detailed information about printer specifications and functionality.