# Gating Revisited: Deep Multi-layer RNNs That Can Be Trained

arXiv: https://arxiv.org/abs/1911.11033 · HTML: https://mlsanity.com/paper/1911.11033/

## Key facts

- Cited by 15 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 25 Nov 2019.

## Built on by (most cited first, 15 of 15)

- [Crop mapping from image time series: deep learning with multi-scale label hierarchies](https://mlsanity.com/paper/2102.08820/) (2021)
- [A Sentinel-2 multi-year, multi-country benchmark dataset for crop classification and segmentation with deep learning](https://mlsanity.com/paper/2204.00951/) (2022)
- [Deep Learning for Satellite Image Time Series Analysis: A Review](https://mlsanity.com/paper/2404.03936/) (2024)
- [Country-wide Retrieval of Forest Structure From Optical and SAR Satellite Imagery With Deep Ensembles](https://mlsanity.com/paper/2111.13154/) (2021)
- [U-TILISE: A Sequence-to-sequence Model for Cloud Removal in Optical Satellite Time Series](https://mlsanity.com/paper/2305.13277/) (2023)
- [AiTLAS: Artificial Intelligence Toolbox for Earth Observation](https://mlsanity.com/paper/2201.08789/) (2022)
- [Deep Loss Convexification for Learning Iterative Models](https://mlsanity.com/paper/2411.10649/) (2024)
- [Mathematical Formalism for Memory Compression in Selective State Space Models](https://mlsanity.com/paper/2410.03158/) (2024)
- [Time-Scale Coupling Between States and Parameters in Recurrent Neural Networks](https://mlsanity.com/paper/2508.12121/) (2025)
- [maskGRU: Tracking Small Objects in the Presence of Large Background Motions](https://mlsanity.com/paper/2201.00467/) (2022)
- [Uncertainty-Aware Deep Attention Recurrent Neural Network for Heterogeneous Time Series Imputation](https://mlsanity.com/paper/2401.02258/) (2024)
- [BILLNET: A Binarized Conv3D-LSTM Network with Logic-gated residual architecture for hardware-efficient video inference](https://mlsanity.com/paper/2501.14495/) (2025)
- [Deep Learning and Earth Observation to Support the Sustainable Development Goals](https://mlsanity.com/paper/2112.11367/) (2021)
- [Correlation recurrent units: A novel neural architecture for improving the predictive performance of time-series data](https://mlsanity.com/paper/2211.16653/) (2022)
- [Cross-Technology Interference: Detection, Avoidance, and Coexistence Mechanisms in the ISM Bands](https://mlsanity.com/paper/2501.06446/) (2025)

## Abstract (arXiv)

We propose a new STAckable Recurrent cell (STAR) for recurrent neural networks (RNNs), which has fewer parameters than widely used LSTM and GRU while being more robust against vanishing or exploding gradients. Stacking recurrent units into deep architectures suffers from two major limitations: (i) many recurrent cells (e.g., LSTMs) are costly in terms of parameters and computation resources; and (ii) deep RNNs are prone to vanishing or exploding gradients during training. We investigate the training of multi-layer RNNs and examine the magnitude of the gradients as they propagate through the network in the "vertical" direction. We show that, depending on the structure of the basic recurrent unit, the gradients are systematically attenuated or amplified. Based on our analysis we design a new type of gated cell that better preserves gradient magnitude. We validate our design on a large number of sequence modelling tasks and demonstrate that the proposed STAR cell allows to build and train deeper recurrent architectures, ultimately leading to improved performance while being computationally more efficient.