Applying LSTM to Time Series Predictable through Time-Window Approaches

By Felix Alexander Gers, Douglas Eck, J. Urgen Schmidhuber. 2000

489 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

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

In Artificial Intelligence

In Signal Processing

Citations in the corpus

DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks (2017, methodology): “We assume that our model distribution Q Θ ( z i,t 0 : T | z i, 1: t 0 − 1 , x i, 1: T ) consists of a product of likelihood factors parametrized by the output h i,t of an autoregressive recurrent network where h is a function implemented by a multi-layer recurrent neural network with LSTM cells.”
Deep Learning With Spatiotemporal Attention-Based LSTM for Industrial Soft Sensor Model Development (2020, background): “This leads to its successful applications in many ﬁelds such as language modeling [36], time series prediction [37], and automatic speech recognition [38].”
Predicting the Price of Bitcoin Using Machine Learning (2018, methodology): “[22] implemented a LSTM for a time series prediction task finding that the LSTM performed as well as the RNN for this task.”
Multimodal sensory fusion for soccer robot self-localization based on long short-term memory recurrent neural network (2017, background): “Our model deals with fixed length of a context window which is 100 rather than the variable length of whole sequences (Gers et al. 2001).”
Combining the real-time wavelet denoising and long-short-term-memory neural network for predicting stock indexes (2017, background): “Long-short-term-memory neural network (LSTM), one of powerful architectures of Recurrent Neural Network (RNN), has been shown successful applications in many engineering domains [6-8].”
The''echo state''approach to analysing and training recurrent neural networks (2001, background): “Another good model, not contained in the survey of [6], was achieved in [12].”
DeepAnT: A Deep Learning Approach for Unsupervised Anomaly Detection in Time Series (2019, background): “It is shown in literature [42], [43] that LSTM performs well on temporal data due to its capability to extract long-term trends in the encountered time series.”
Efficient Orthogonal Parametrisation of Recurrent Neural Networks Using Householder Reflections (2016, methodology): “We tested our model on the copy task (Gers et al., 2001) described in details in (Gers et al., 2001; Arjovsky et al., 2015).”
Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks (2018, methodology): “The LSTM is trained to forecast the derivative of the reduced order state dzt/dt as in [34].”
Audio Super Resolution using Neural Networks (2017, background): “In the machine learning literature, time series signals have most often been modeled with auto-regressive models, of which variants of recurrent networks are a special case (Gers et al., 2001; Maas et al., 2012; Mehri et al., 2016).”
MOrdReD: Memory-based Ordinal Regression Deep Neural Networks for Time Series Forecasting (2018, background): “Indeed, such networks could only be trained at small scale until recently, and this could be a potential reason why early ﬁndings showed that they were easily outperformed by simpler, memory-less methods [21] in the context of time series.”
Deep forest (2017, methodology): “In addition to an MLP with input–1024–512–output structure, we also evaluate a recurrent neural network, LSTM [27], with 128 hidden units and a sequence length of 6 (500-dim input vector per second).”
Position-Based Content Attention for Time Series Forecasting with Sequence-to-Sequence RNNs (2017, background): “They have furthermore been shown to outperform traditional RNNs on various temporal tasks [23,24].”
Distributed Multi-task Learning for Sensor Network (2017, methodology): “For example, the approaches based on RNN with LSTM [11, 9, 13] may have better prediction performance on non-stationary time series data.”
Multi-period Time Series Modeling with Sparsity via Bayesian Variational Inference (2017, methodology): “Especially LSTM model cannot work well on cases where the prediction is primarily based on recent past observations [5].”