Architectural Complexity Measures of Recurrent Neural Networks

In this paper, we systematically analyze the connecting architectures of recurrent neural networks (RNNs). Our main contribution is twofold: first, we present a rigorous graph-theoretic framework describing the connecting architectures of RNNs in general. Second, we propose three architecture complexity measures of RNNs: (a) the recurrent depth, which captures the RNN's over-time nonlinear complexity, (b) the feedforward depth, which captures the local input-output nonlinearity (similar to the "depth" in feedforward neural networks (FNNs)), and (c) the recurrent skip coefficient which captures how rapidly the information propagates over time. We rigorously prove each measure's existence and computability. Our experimental results show that RNNs might benefit from larger recurrent depth and feedforward depth. We further demonstrate that increasing recurrent skip coefficient offers performance boosts on long term dependency problems.

PDF Abstract NeurIPS 2016 PDF NeurIPS 2016 Abstract
No code implementations yet. Submit your code now


Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Language Modelling Text8 td-LSTM (Zhang et al., 2016) Bit per Character (BPC) 1.63 # 24
Language Modelling Text8 td-LSTM-large Bit per Character (BPC) 1.49 # 23


No methods listed for this paper. Add relevant methods here