Flattened Convolutional Neural Networks for Feedforward Acceleration
This work addresses efficiency issues in neural network inference for applications requiring fast feedforward passes, though it is incremental as it builds on existing low-rank filter methods.
The paper tackles the problem of parameter redundancy in convolutional neural networks by introducing flattened networks with one-dimensional filters, achieving comparable accuracy to conventional models while providing around two times speed-up in feedforward execution.
We present flattened convolutional neural networks that are designed for fast feedforward execution. The redundancy of the parameters, especially weights of the convolutional filters in convolutional neural networks has been extensively studied and different heuristics have been proposed to construct a low rank basis of the filters after training. In this work, we train flattened networks that consist of consecutive sequence of one-dimensional filters across all directions in 3D space to obtain comparable performance as conventional convolutional networks. We tested flattened model on different datasets and found that the flattened layer can effectively substitute for the 3D filters without loss of accuracy. The flattened convolution pipelines provide around two times speed-up during feedforward pass compared to the baseline model due to the significant reduction of learning parameters. Furthermore, the proposed method does not require efforts in manual tuning or post processing once the model is trained.