Low Rank and Sparse Fourier Structure in Recurrent Networks Trained on Modular Addition
This provides incremental insights into neural network mechanisms for researchers studying empirical phenomena like grokking in deep learning.
The paper tackled the problem of understanding how Recurrent Neural Networks (RNNs) solve modular addition tasks, showing that they use a Fourier Multiplication strategy with low-rank, sparse Fourier structures, and performance degrades drastically when multiple frequencies are removed.
Modular addition tasks serve as a useful test bed for observing empirical phenomena in deep learning, including the phenomenon of \emph{grokking}. Prior work has shown that one-layer transformer architectures learn Fourier Multiplication circuits to solve modular addition tasks. In this paper, we show that Recurrent Neural Networks (RNNs) trained on modular addition tasks also use a Fourier Multiplication strategy. We identify low rank structures in the model weights, and attribute model components to specific Fourier frequencies, resulting in a sparse representation in the Fourier space. We also show empirically that the RNN is robust to removing individual frequencies, while the performance degrades drastically as more frequencies are ablated from the model.