TCN: Temporal Convolutional Network

This is a generic Temporal Convolutional Network implementation as proposed in An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling by Bai et al. developed with Tensorflow.

The implementation has been tested on the following task contained in this repository:

• Sequential MNIST
• Copy memory
• Text classification on the IMDB movie dataset

Architectural elements in a TCN. (a) A dilated causal convolution with dilation factors d = 1, 2, 4 and filter size k = 3. (b) TCN residual block. (c) An example of residual connection in a TCN

The corresponding elements in TensorBoard