: A Long Short-Term

By Nandini Geddlehally, K. Gopala, Kishore Gopala Rao

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Model Architectures

Citations in the corpus

Benchmarking Deep Reinforcement Learning for Continuous Control (2016, positive): “For all partially observable tasks, we use a recurrent neural network with a single hidden layer consisting of 32 LSTM hidden units (Hochreiter & Schmidhuber, 1997).”
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention (2015, neutral): “We use a long short-term memory (LSTM) network (Hochreiter & Schmidhuber, 1997) that produces a caption by generating one word at every time step conditioned on a context vector, the previous hidden state and the previously generated words.”
Deep Reinforcement Learning amidst Continual Structured Non-Stationarity (2021, unknown): “Finally, we approximate p ( z i | z 1: i − 1 ) and p ( z i | τ 1: i − 1 ) with a shared long short-term memory (LSTM) network (Hochreiter & Schmidhuber, 1997), which receives z i − 1 from q ( z i − 1 | τ i − 1 ) and hidden state h i − 1 , and produces z i and the next hidden state h i .”
An Experimental Review on Deep Learning Architectures for Time Series Forecasting (2020, unknown): “In 1997, Long Short-Term Memory (LSTM) networks 56 were introduced as a solution for ERNN’s problems.”
Gated Linear Attention Transformers with Hardware-Efficient Training (2023, unknown): “1 does not have a decay term or a forget gate, which has been shown to be crucial in RNNs (Hochreiter & Schmidhuber, 1997; Cho et al., 2014; van der Westhuizen & Lasenby, 2018).”
Self-Attention Graph Pooling (2019, unknown): “We assume that the readout layer is unnecessary because the LSTM module (Defferrard et al., 2016), GCNConv (Kipf & Welling, 2016), SAGEConv (Hamilton et al., 2017), and GATConv(heads=6) (Velikovi et al., 2018). produces embeddings for graphs invariant to the order of nodes.”