NeuMiss

NeuMiss is a neural network architecture aimed at handling missing values, usually used as a preprocessing layer.

For a detailed description of the problem of encoding dirty categorical data, see NeuMiss networks: differentiable programming for supervised learning with missing values¹ and What’s a good imputation to predict with missing values?².

References

---

1. Marine Le Morvan, et al. NeuMiss networks: differentiable programming for supervised learning with missing values. 2020. Advances in Neural Information Processing Systems 33 (NeurIPS 2020)↩

2. Marine Le Morvan, et al. What’s a good imputation to predict with missing values. 2021.↩