# Anti-Distillation: Improving reproducibility of deep networks

arXiv: https://arxiv.org/abs/2010.09923 · HTML: https://mlsanity.com/paper/2010.09923/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 19 Oct 2020.

## Built on by (most cited first, 13 of 13)

- [Similarity of Neural Network Models: A Survey of Functional and Representational Measures](https://mlsanity.com/paper/2305.06329/) (2023)
- [Randomness In Neural Network Training: Characterizing The Impact of Tooling](https://mlsanity.com/paper/2106.11872/) (2021)
- [On the Factory Floor: ML Engineering for Industrial-Scale Ads Recommendation Models](https://mlsanity.com/paper/2209.05310/) (2022)
- [Reproducibility in Optimization: Theoretical Framework and Limits](https://mlsanity.com/paper/2202.04598/) (2022)
- [FAIR-Ensemble: When Fairness Naturally Emerges From Deep Ensembling](https://mlsanity.com/paper/2303.00586/) (2023)
- [Synthesizing Irreproducibility in Deep Networks](https://mlsanity.com/paper/2102.10696/) (2021)
- [Reproducibility of Machine Learning: Terminology, Recommendations and Open Issues](https://mlsanity.com/paper/2302.12691/) (2023)
- [Reducing Model Jitter: Stable Re-training of Semantic Parsers in Production Environments](https://mlsanity.com/paper/2204.04735/) (2022)
- [On the Prediction Instability of Graph Neural Networks](https://mlsanity.com/paper/2205.10070/) (2022)
- [Handling many conversions per click in modeling delayed feedback](https://mlsanity.com/paper/2101.02284/) (2021)
- [Optimal Guarantees for Algorithmic Reproducibility and Gradient Complexity in Convex Optimization](https://mlsanity.com/paper/2310.17759/) (2023)
- [Real World Large Scale Recommendation Systems Reproducibility and Smooth Activations](https://mlsanity.com/paper/2202.06499/) (2022)
- [Dropout Prediction Uncertainty Estimation Using Neuron Activation Strength](https://mlsanity.com/paper/2110.06435/) (2021)

## Abstract (arXiv)

Deep networks have been revolutionary in improving performance of machine learning and artificial intelligence systems. Their high prediction accuracy, however, comes at a price of model irreproducibility/ in very high levels that do not occur with classical linear models. Two models, even if they are supposedly identical, with identical architecture and identical trained parameter sets, and that are trained on the same set of training examples, while possibly providing identical average prediction accuracies, may predict very differently on individual, previously unseen, examples. Prediction differences/ may be as large as the order of magnitude of the predictions themselves. Ensembles have been shown to somewhat mitigate this behavior, but without an extra push, may not be utilizing their full potential. In this work, a novel approach, Anti-Distillation/, is proposed to address irreproducibility in deep networks, where ensemble models are used to generate predictions. Anti-Distillation forces ensemble components away from one another by techniques like de-correlating their outputs over mini-batches of examples, forcing them to become even more different and more diverse. Doing so enhances the benefit of ensembles, making the final predictions more reproducible. Empirical results demonstrate substantial prediction difference reductions achieved by Anti-Distillation on benchmark and real datasets.