Interpretable Neuron Structuring with Graph Spectral Regularization.
Interpretable Neuron Structuring with Graph Spectral Regularization.
复制标题
DOI:
10.1007/978-3-030-44584-3_40
复制
发表时间:
2020-04
期刊:
影响因子:
--
通讯作者:
Krishnaswamy S
中科院分区:
文献类型:
--
作者:
Tong A;van Dijk D;Stanley JS 3rd;Amodio M;Yim K;Muhle R;Noonan J;Wolf G;Krishnaswamy S
While neural networks are powerful approximators used to classify or embed data into lower dimensional spaces, they are often regarded as black boxes with uninterpretable features. Here we propose Graph Spectral Regularization for making hidden layers more interpretable without significantly impacting performance on the primary task. Taking inspiration from spatial organization and localization of neuron activations in biological networks, we use a graph Laplacian penalty to structure the activations within a layer. This penalty encourages activations to be smooth either on a predetermined graph or on a feature-space graph learned from the data via co-activations of a hidden layer of the neural network. We show numerous uses for this additional structure including cluster indication and visualization in biological and image data sets.
登录
查看更多内容
影响因子:
46.9
作者:
Setty M;Tadmor MD;Reich-Zeliger S;Angel O;Salame TM;Kathail P;Choi K;Bendall S;Friedman N;Pe'er D
通讯作者:
Pe'er D
影响因子:
--
作者:
Scarselli, Franco;Gori, Marco;Monfardini, Gabriele
通讯作者:
Monfardini, Gabriele
DOI:
10.1007/978-3-540-27819-1_43
发表时间:
2004-01-01
期刊:
LEARNING THEORY, PROCEEDINGS
影响因子:
--
作者:
Belkin, M;Matveeva, I;Niyogi, P
通讯作者:
Niyogi, P
DOI:
10.1109/tnnls.2020.2978386
发表时间:
2021-01-01
影响因子:
10.4
作者:
Wu, Zonghan;Pan, Shirui;Yu, Philip S.
通讯作者:
Yu, Philip S.
影响因子:
2.9
作者:
LeCun, Y.;Boser, B.;Jackel, L. D.
通讯作者:
Jackel, L. D.