Predictive Coding Deep Neural Network in Chainer (ver.tb8f)

The PredNet*, written in Keras, was ported to Chainer and reconstructed for convenience.Visualization of internal images in the network models was performed using a self-made converter program and Tesorboard in Tensorflow.<br><br>*Lotter, W., Kreiman, G., &amp; Cox, D., Deep predictive coding networks for video prediction and unsupervised learning, arXiv 1605.08104.<br>This program is new version used inWatanabe E, Kitaoka A, Sakamoto K, Yasugi M and Tanaka K, Illusory Motion Reproduced by Deep Neural Networks Trained for Prediction. Frontiers in Psychology 9:345. doi: 10.3389/fpsyg.2018.00345 (2018)

本研究所用的PredNet*最初基于Keras编写，为便于使用已被移植至Chainer框架并完成重构。我们通过自研转换程序与TensorFlow中的Tesorboard完成了网络模型内部特征图的可视化工作。

*Lotter, W.、Kreiman, G. 与 Cox, D. 所著《用于视频预测与无监督学习的深度预测编码网络》，arXiv 1605.08104。
本程序为Watanabe E、Kitaoka A、Sakamoto K、Yasugi M及Tanaka K等人在2018年发表于《Frontiers in Psychology》第9卷第345篇文章的论文《基于预测任务训练的深度神经网络重现视错觉运动》（DOI: 10.3389/fpsyg.2018.00345）中所用版本的更新版。