Sequence Modeling with Multiresolution Convolutional Memory

Jiaxin Shi, Ke Alexander Wang, Emily Fox
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:31312-31327, 2023.

Abstract

Efficiently capturing the long-range patterns in sequential data sources salient to a given task—such as classification and generative modeling—poses a fundamental challenge. Popular approaches in the space tradeoff between the memory burden of brute-force enumeration and comparison, as in transformers, the computational burden of complicated sequential dependencies, as in recurrent neural networks, or the parameter burden of convolutional networks with many or large filters. We instead take inspiration from wavelet-based multiresolution analysis to define a new building block for sequence modeling, which we call a MultiresLayer. The key component of our model is the multiresolution convolution, capturing multiscale trends in the input sequence. Our MultiresConv can be implemented with shared filters across a dilated causal convolution tree. Thus it garners the computational advantages of convolutional networks and the principled theoretical motivation of wavelet decompositions. Our MultiresLayer is straightforward to implement, requires significantly fewer parameters, and maintains at most a $O(N \log N)$ memory footprint for a length $N$ sequence. Yet, by stacking such layers, our model yields state-of-the-art performance on a number of sequence classification and autoregressive density estimation tasks using CIFAR-10, ListOps, and PTB-XL datasets.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-shi23f, title = {Sequence Modeling with Multiresolution Convolutional Memory}, author = {Shi, Jiaxin and Wang, Ke Alexander and Fox, Emily}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {31312--31327}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/shi23f/shi23f.pdf}, url = {https://proceedings.mlr.press/v202/shi23f.html}, abstract = {Efficiently capturing the long-range patterns in sequential data sources salient to a given task—such as classification and generative modeling—poses a fundamental challenge. Popular approaches in the space tradeoff between the memory burden of brute-force enumeration and comparison, as in transformers, the computational burden of complicated sequential dependencies, as in recurrent neural networks, or the parameter burden of convolutional networks with many or large filters. We instead take inspiration from wavelet-based multiresolution analysis to define a new building block for sequence modeling, which we call a MultiresLayer. The key component of our model is the multiresolution convolution, capturing multiscale trends in the input sequence. Our MultiresConv can be implemented with shared filters across a dilated causal convolution tree. Thus it garners the computational advantages of convolutional networks and the principled theoretical motivation of wavelet decompositions. Our MultiresLayer is straightforward to implement, requires significantly fewer parameters, and maintains at most a $O(N \log N)$ memory footprint for a length $N$ sequence. Yet, by stacking such layers, our model yields state-of-the-art performance on a number of sequence classification and autoregressive density estimation tasks using CIFAR-10, ListOps, and PTB-XL datasets.} }
Endnote
%0 Conference Paper %T Sequence Modeling with Multiresolution Convolutional Memory %A Jiaxin Shi %A Ke Alexander Wang %A Emily Fox %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-shi23f %I PMLR %P 31312--31327 %U https://proceedings.mlr.press/v202/shi23f.html %V 202 %X Efficiently capturing the long-range patterns in sequential data sources salient to a given task—such as classification and generative modeling—poses a fundamental challenge. Popular approaches in the space tradeoff between the memory burden of brute-force enumeration and comparison, as in transformers, the computational burden of complicated sequential dependencies, as in recurrent neural networks, or the parameter burden of convolutional networks with many or large filters. We instead take inspiration from wavelet-based multiresolution analysis to define a new building block for sequence modeling, which we call a MultiresLayer. The key component of our model is the multiresolution convolution, capturing multiscale trends in the input sequence. Our MultiresConv can be implemented with shared filters across a dilated causal convolution tree. Thus it garners the computational advantages of convolutional networks and the principled theoretical motivation of wavelet decompositions. Our MultiresLayer is straightforward to implement, requires significantly fewer parameters, and maintains at most a $O(N \log N)$ memory footprint for a length $N$ sequence. Yet, by stacking such layers, our model yields state-of-the-art performance on a number of sequence classification and autoregressive density estimation tasks using CIFAR-10, ListOps, and PTB-XL datasets.
APA
Shi, J., Wang, K.A. & Fox, E.. (2023). Sequence Modeling with Multiresolution Convolutional Memory. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:31312-31327 Available from https://proceedings.mlr.press/v202/shi23f.html.

Related Material