# N2N Learning: Network to Network Compression via Policy Gradient Reinforcement Learning

arXiv: https://arxiv.org/abs/1709.06030 · HTML: https://mlsanity.com/paper/1709.06030/

## Key facts

- Cited by 33 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Sep 2017.

## Built on by (most cited first, 30 of 33)

- [Low-rank lottery tickets: finding efficient low-rank neural networks via matrix differential equations](https://mlsanity.com/paper/2205.13571/) (2022)
- [SPDY: Accurate Pruning with Speedup Guarantees](https://mlsanity.com/paper/2201.13096/) (2022)
- [Visual Tuning](https://mlsanity.com/paper/2305.06061/) (2023)
- [DECORE: Deep Compression with Reinforcement Learning](https://mlsanity.com/paper/2106.06091/) (2021)
- [EvoPress: Accurate Dynamic Model Compression via Evolutionary Search](https://mlsanity.com/paper/2410.14649/) (2024)
- [Is Complexity Required for Neural Network Pruning? A Case Study on Global Magnitude Pruning](https://mlsanity.com/paper/2209.14624/) (2022)
- [Robust low-rank training via approximate orthonormal constraints](https://mlsanity.com/paper/2306.01485/) (2023)
- [OpenRFT: Adapting Reasoning Foundation Model for Domain-specific Tasks with Reinforcement Fine-Tuning](https://mlsanity.com/paper/2412.16849/) (2024)
- [Automatic Attention Pruning: Improving and Automating Model Pruning using Attentions](https://mlsanity.com/paper/2303.08595/) (2023)
- [Reconsidering CO2 emissions from Computer Vision](https://mlsanity.com/paper/2104.08702/) (2021)
- [Jointly Training and Pruning CNNs via Learnable Agent Guidance and Alignment](https://mlsanity.com/paper/2403.19490/) (2024)
- [Self-paced Resistance Learning against Overfitting on Noisy Labels](https://mlsanity.com/paper/2105.03059/) (2021)
- [Combining Multi-Objective Bayesian Optimization with Reinforcement Learning for TinyML](https://mlsanity.com/paper/2305.14109/) (2023)
- [Enhanced Gradient for Differentiable Architecture Search](https://mlsanity.com/paper/2103.12529/) (2021)
- [Knowledge Amalgamation for Object Detection with Transformers](https://mlsanity.com/paper/2203.03187/) (2022)
- [Adaptive Distillation: Aggregating Knowledge from Multiple Paths for Efficient Distillation](https://mlsanity.com/paper/2110.09674/) (2021)
- [Adaptive Activation-based Structured Pruning](https://mlsanity.com/paper/2201.10520/) (2022)
- [Divide-and-Conquer the NAS puzzle in Resource Constrained Federated Learning Systems](https://mlsanity.com/paper/2305.07135/) (2023)
- [Resource Efficient Neural Networks Using Hessian Based Pruning](https://mlsanity.com/paper/2306.07030/) (2023)
- [Communication-Computation Efficient Device-Edge Co-Inference via AutoML](https://mlsanity.com/paper/2108.13009/) (2021)
- [Knowledge Distillation: Enhancing Neural Network Compression with Integrated Gradients](https://mlsanity.com/paper/2503.13008/) (2025)
- [One-for-All Pruning: A Universal Model for Customized Compression of Large Language Models](https://mlsanity.com/paper/2505.12216/) (2025)
- [SONATA: Self-adaptive Evolutionary Framework for Hardware-aware Neural Architecture Search](https://mlsanity.com/paper/2402.13204/) (2024)
- [AutoDFP: Automatic Data-Free Pruning via Channel Similarity Reconstruction](https://mlsanity.com/paper/2403.08204/) (2024)
- [Mining the Weights Knowledge for Optimizing Neural Network Structures](https://mlsanity.com/paper/2110.05954/) (2021)
- [Conditional Generative Data-free Knowledge Distillation](https://mlsanity.com/paper/2112.15358/) (2021)
- [Model Blending for Text Classification](https://mlsanity.com/paper/2208.02819/) (2022)
- [NAS-based Recursive Stage Partial Network (RSPNet) for Light-Weight Semantic Segmentation](https://mlsanity.com/paper/2210.00698/) (2022)
- [Edge-Cloud Cooperation for DNN Inference via Reinforcement Learning and Supervised Learning](https://mlsanity.com/paper/2210.05182/) (2022)
- [Reinforced Compressive Neural Architecture Search for Versatile Adversarial Robustness](https://mlsanity.com/paper/2406.06792/) (2024)

## Abstract (arXiv)

While bigger and deeper neural network architectures continue to advance the state-of-the-art for many computer vision tasks, real-world adoption of these networks is impeded by hardware and speed constraints. Conventional model compression methods attempt to address this problem by modifying the architecture manually or using pre-defined heuristics. Since the space of all reduced architectures is very large, modifying the architecture of a deep neural network in this way is a difficult task. In this paper, we tackle this issue by introducing a principled method for learning reduced network architectures in a data-driven way using reinforcement learning. Our approach takes a larger `teacher' network as input and outputs a compressed `student' network derived from the `teacher' network. In the first stage of our method, a recurrent policy network aggressively removes layers from the large `teacher' model. In the second stage, another recurrent policy network carefully reduces the size of each remaining layer. The resulting network is then evaluated to obtain a reward -- a score based on the accuracy and compression of the network. Our approach uses this reward signal with policy gradients to train the policies to find a locally optimal student network. Our experiments show that we can achieve compression rates of more than 10x for models such as ResNet-34 while maintaining similar performance to the input `teacher' network. We also present a valuable transfer learning result which shows that policies which are pre-trained on smaller `teacher' networks can be used to rapidly speed up training on larger `teacher' networks.