Analysis for
Deep Learning

We analyzed a trained network to extract features which mainly contribute the inference. Based on the analysis, we developed a simple solution to generate explanations of the inference processes

.

Improving Transparency of Deep Neural Inference Process


Hiroshi Kuwajima, Masayuki Tanaka and Masatoshi Okutomi
Progress in Artificial Intelligence, Vol.8, Issue2, pp.273-285, June, 2019

[Springer SharedIt ]

深層学習における推論根拠解析


桑島洋, 田中正行
第24回画像センシングシンポジウム（SSII2018）, pp.SO2-IS2-02-1-8, June, 2018.

[オーディエンス賞]

Network Analysis for Explanation


Hiroshi Kuwajima, Masayuki Tanaka
Transparent and Interpretable Machine Learning in Safety Critical Environments(NIPS2017 workshop), December, 2017.

[arxiv] [poster]

Related project: Deep Learning (General Framework)
Return to Research Projects