Exploring Interpretable LSTM Neural Networks over Multi-Variable Data

By Tian Guo, Tao Lin, Nino Antulov-Fantulin. 2019

184 citations recorded by the source corpus.

For recurrent neural networks trained on time series with target and exogenous variables, in addition to accurate prediction, it is also desired to provide interpretable insights into the data. In this paper, we explore the structure of LSTM recurrent neural networks to learn variable-wise hidden states, with the aim to capture different dynamics in multi-variable time series and distinguish the contribution of variables to the prediction. With these variable-wise hidden states, a mixture attention mechanism is proposed to model the generative process of the target. Then we develop associated training methods to jointly learn network parameters, variable and temporal importance w.r.t the prediction of the target variable. Extensive experiments on real datasets demonstrate enhanced prediction performance by capturing the dynamics of different variables. Meanwhile, we evaluate the interpretation results both qualitatively and quantitatively. It exhibits the prospect as an end-to-end framework for both forecasting and knowledge extraction over multi-variable data.

Subfields: Artificial Intelligence, Management Science and Operations Research, 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

1. Cho-Jui Hsieh
2. Hongge Chen
3. Duane S. Boning
4. Bryan Lim
5. Si Si
6. Nicolas Loeff
7. Sanjiv Kumar
8. Tomas Pfister

In Management Science and Operations Research

1. Tomas Pfister
2. Sercan Ö. Arik
3. Ian Covert
4. Ahmed Hosny
5. Bryan Lim
6. Hugo J. W. L. Aerts
7. Sanjiv Kumar
8. Cho-Jui Hsieh

In Signal Processing

1. Ali Shojaie
2. Emily B. Fox
3. Tomas Pfister
4. Sercan Ö. Arik
5. Si Si
6. Ian Covert
7. Ahmed Hosny
8. Bryan Lim

Citations in the corpus

• Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting (2019, methodology): “For example, Interpretable Multi-Variable LSTMs [18] partition the hidden state such that each variable contributes uniquely to its own memory segment, and weights memory segments to determine variable contributions.”
• Explainable CNN-attention Networks (C-Attention Network) for Automated Detection of Alzheimer's Disease (2020, background): “The structure of LSTM was explored and contribution of variables to the prediction in multi-variable time series was captured in [20].”
• Neural Granger Causality (2021, methodology): “The IMV-LSTM [56], [57] uses attention weights to provide greater interpretability than standard LSTMs, and it detects Granger causal relationships by aggregating its attention weights.”
• Artificial Intelligence for Clinical Oncology (2021, methodology): “…a growing research field dedicated to investigation of interpretability issues, and several techniques, such as saliency maps, hidden-states analysis, variable importance metrics, and feature visualizations can illuminate some aspects of AI prediction rationale (Guo et al., 2019; Olah et al., 2018).”
• Multi-Stage Influence Function (2020, background): “Feature based approaches aim to explain predictions with respect to model variables, and trace back the contribution of variables to the prediction [17, 9, 21, 23, 22, 25, 8, 6, 1].”