Continuous-in-Depth Neural Networks

• URL: http://arxiv.org/abs/2008.02389v1
• Date: Wed, 5 Aug 2020 22:54:09 GMT
• Title: Continuous-in-Depth Neural Networks
• Authors: Alejandro F. Queiruga, N. Benjamin Erichson, Dane Taylor and Michael W. Mahoney
• Abstract summary: We first show that ResNets fail to be meaningful dynamical in this richer sense. We then demonstrate that neural network models can learn to represent continuous dynamical systems. We introduce ContinuousNet as a continuous-in-depth generalization of ResNet architectures.
• Score: 107.47887213490134
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent work has attempted to interpret residual networks (ResNets) as one step of a forward Euler discretization of an ordinary differential equation, focusing mainly on syntactic algebraic similarities between the two systems. Discrete dynamical integrators of continuous dynamical systems, however, have a much richer structure. We first show that ResNets fail to be meaningful dynamical integrators in this richer sense. We then demonstrate that neural network models can learn to represent continuous dynamical systems, with this richer structure and properties, by embedding them into higher-order numerical integration schemes, such as the Runge Kutta schemes. Based on these insights, we introduce ContinuousNet as a continuous-in-depth generalization of ResNet architectures. ContinuousNets exhibit an invariance to the particular computational graph manifestation. That is, the continuous-in-depth model can be evaluated with different discrete time step sizes, which changes the number of layers, and different numerical integration schemes, which changes the graph connectivity. We show that this can be used to develop an incremental-in-depth training scheme that improves model quality, while significantly decreasing training time. We also show that, once trained, the number of units in the computational graph can even be decreased, for faster inference with little-to-no accuracy drop.

Related papers

• How neural networks learn to classify chaotic time series [77.34726150561087]
  We study the inner workings of neural networks trained to classify regular-versus-chaotic time series. We find that the relation between input periodicity and activation periodicity is key for the performance of LKCNN models.
  arXiv  Detail & Related papers  (2023-06-04T08:53:27Z)
• Piecewise-Velocity Model for Learning Continuous-time Dynamic Node Representations [0.0]
  Piecewise-Veable Model (PiVeM) for representation of continuous-time dynamic networks. We show that PiVeM can successfully represent network structure and dynamics in ultra-low two-dimensional spaces. It outperforms relevant state-of-art methods in downstream tasks such as link prediction.
  arXiv  Detail & Related papers  (2022-12-23T13:57:56Z)
• The Underlying Correlated Dynamics in Neural Training [6.385006149689549]
  Training of neural networks is a computationally intensive task. We propose a model based on the correlation of the parameters' dynamics, which dramatically reduces the dimensionality. This representation enhances the understanding of the underlying training dynamics and can pave the way for designing better acceleration techniques.
  arXiv  Detail & Related papers  (2022-12-18T08:34:11Z)
• Dynamic Graph Message Passing Networks for Visual Recognition [112.49513303433606]
  Modelling long-range dependencies is critical for scene understanding tasks in computer vision. A fully-connected graph is beneficial for such modelling, but its computational overhead is prohibitive. We propose a dynamic graph message passing network, that significantly reduces the computational complexity.
  arXiv  Detail & Related papers  (2022-09-20T14:41:37Z)
• Efficient-Dyn: Dynamic Graph Representation Learning via Event-based Temporal Sparse Attention Network [2.0047096160313456]
  Dynamic graph neural networks have received more and more attention from researchers. We propose a novel dynamic graph neural network, Efficient-Dyn. It adaptively encodes temporal information into a sequence of patches with an equal amount of temporal-topological structure.
  arXiv  Detail & Related papers  (2022-01-04T23:52:24Z)
• Spectral Transform Forms Scalable Transformer [1.19071399645846]
  This work learns from the philosophy of self-attention and proposes an efficient spectral-based neural unit that employs informative long-range temporal interaction. The developed spectral window unit (SW) model predicts scalable dynamic graphs with assured efficiency.
  arXiv  Detail & Related papers  (2021-11-15T08:46:01Z)
• Graph-Based Neural Network Models with Multiple Self-Supervised Auxiliary Tasks [79.28094304325116]
  Graph Convolutional Networks are among the most promising approaches for capturing relationships among structured data points. We propose three novel self-supervised auxiliary tasks to train graph-based neural network models in a multi-task fashion.
  arXiv  Detail & Related papers  (2020-11-14T11:09:51Z)
• Multipole Graph Neural Operator for Parametric Partial Differential Equations [57.90284928158383]
  One of the main challenges in using deep learning-based methods for simulating physical systems is formulating physics-based data. We propose a novel multi-level graph neural network framework that captures interaction at all ranges with only linear complexity. Experiments confirm our multi-graph network learns discretization-invariant solution operators to PDEs and can be evaluated in linear time.
  arXiv  Detail & Related papers  (2020-06-16T21:56:22Z)
• Liquid Time-constant Networks [117.57116214802504]
  We introduce a new class of time-continuous recurrent neural network models. Instead of declaring a learning system's dynamics by implicit nonlinearities, we construct networks of linear first-order dynamical systems. These neural networks exhibit stable and bounded behavior, yield superior expressivity within the family of neural ordinary differential equations.
  arXiv  Detail & Related papers  (2020-06-08T09:53:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.