# Increasing the Interpretability of Recurrent Neural Networks Using Hidden Markov Models

arXiv: https://arxiv.org/abs/1606.05320 · HTML: https://mlsanity.com/paper/1606.05320/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Jun 2016.

## Built on by (most cited first, 10 of 10)

- [M2Lens: Visualizing and Explaining Multimodal Models for Sentiment Analysis](https://mlsanity.com/paper/2107.08264/) (2021)
- [Generative learning for nonlinear dynamics](https://mlsanity.com/paper/2311.04128/) (2023)
- [A Review of Multimodal Explainable Artificial Intelligence: Past, Present and Future](https://mlsanity.com/paper/2412.14056/) (2024)
- [Weighted Automata Extraction and Explanation of Recurrent Neural Networks for Natural Language Tasks](https://mlsanity.com/paper/2306.14040/) (2023)
- [Hybrid hidden Markov LSTM for short-term traffic flow prediction](https://mlsanity.com/paper/2307.04954/) (2023)
- [Absolute Value Constraint: The Reason for Invalid Performance Evaluation Results of Neural Network Models for Stock Price Prediction](https://mlsanity.com/paper/2101.10942/) (2021)
- [BTPK-based interpretable method for NER tasks based on Talmudic Public Announcement Logic](https://mlsanity.com/paper/2201.09523/) (2022)
- [AI for Investment: A Platform Disruption](https://mlsanity.com/paper/2311.06251/) (2023)
- [Latent States in Neural Networks: Recovering the Temporal Structure of Drifting Data from Model Weights](https://mlsanity.com/paper/2607.27482/) (2026)
- [Temporally Interpretable Differentiable Decision Trees](https://mlsanity.com/paper/2610.10367/) (2026)

## Abstract (arXiv)

As deep neural networks continue to revolutionize various application domains, there is increasing interest in making these powerful models more understandable and interpretable, and narrowing down the causes of good and bad predictions. We focus on recurrent neural networks (RNNs), state of the art models in speech recognition and translation. Our approach to increasing interpretability is by combining an RNN with a hidden Markov model (HMM), a simpler and more transparent model. We explore various combinations of RNNs and HMMs: an HMM trained on LSTM states; a hybrid model where an HMM is trained first, then a small LSTM is given HMM state distributions and trained to fill in gaps in the HMM's performance; and a jointly trained hybrid model. We find that the LSTM and HMM learn complementary information about the features in the text.