In this work, we develop convolutional neural generative coding (Conv-NGC), a generalization of predictive coding to the case of convolution/deconvolution-based computation. Specifically, we concretely implement a flexible neurobiologically-motivated algorithm that progressively refines latent state feature maps in order to dynamically form a more accurate internal representation/reconstruction model of natural images. The performance of the resulting sensory processing system is evaluated on complex datasets such as Color-MNIST, CIFAR-10, and Street House View Numbers (SVHN). We study the effectiveness of our brain-inspired model on the tasks of reconstruction and image denoising and find that it is competitive with convolutional auto-encoding systems trained by backpropagation of errors and outperforms them with respect to out-of-distribution reconstruction (including the full 90k CINIC-10 test set).

翻译：本文开发了卷积神经生成编码（Conv-NGC），这是将预测编码推广到基于卷积/反卷积计算的一种方法。具体而言，我们实现了一种灵活的、神经生物学启发的算法，该算法逐步精化潜在状态特征图，以动态构建自然图像的更精确内部表示/重建模型。所得到的感知处理系统的性能在复杂数据集（如Color-MNIST、CIFAR-10和街景门牌号（SVHN））上进行了评估。我们研究了这种脑启发模型在重建和图像去噪任务上的有效性，发现其与通过误差反向传播训练的卷积自编码系统具有竞争力，并且在分布外重建（包括完整的90k CINIC-10测试集）方面优于后者。