Papers › Learning Implicitly Recurrent CNNs Through Parameter Sharing

Learning Implicitly Recurrent CNNs Through Parameter Sharing

26 Feb 2019ICLR 2019 5arXiv:1902.09701archive 2025-07-28

Pedro Savarese, Michael Maire

We introduce a parameter sharing scheme, in which different layers of a convolutional neural network (CNN) are defined by a learned linear combination of parameter tensors from a global bank of templates. Restricting the number of templates yields a flexible hybridization of traditional CNNs and recurrent networks. Compared to traditional CNNs, we demonstrate substantial parameter savings on standard image classification tasks, while maintaining accuracy. Our simple parameter sharing scheme, though defined via soft weights, in practice often yields trained networks with near strict recurrent structure; with negligible side effects, they convert into networks with actual loops. Training these networks thus implicitly involves discovery of suitable recurrent architectures. Though considering only the design aspect of recurrent links, our trained networks achieve accuracy competitive with those built using state-of-the-art neural architecture search (NAS) procedures. Our hybridization of recurrent and convolutional networks may also represent a beneficial architectural bias. Specifically, on synthetic tasks which are algorithmic in nature, our hybrid networks both train faster and extrapolate better to test examples outside the span of the training set.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

lolemacs/soft-sharing officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationNeural Architecture Searchimage-classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 Shared WRN Percentage correct 97.47 #80 of 265 Archive leaderboard report
Image Classification CIFAR-100 Shared WRN Percentage correct 82.57 #102 of 211 Archive leaderboard report
Neural Architecture Search CIFAR-10 Soft Parameter Sharing Search Time (GPU days) 0.7 #22 of 41 Archive leaderboard report
Neural Architecture Search CIFAR-10 Soft Parameter Sharing Top-1 Error Rate 2.53% #22 of 41 Archive leaderboard report
Neural Architecture Search CIFAR-10 Image Classification Soft Parameter Sharing Params 33.5 #12 of 19 Archive leaderboard report
Neural Architecture Search CIFAR-10 Image Classification Soft Parameter Sharing Percentage error 2.53 #12 of 19 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

LSTMSigmoid ActivationSoftmaxTanh Activation

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections