mlsanityData to 7 Oct 2026

Learning to Reverse DNNs from AI Programs Automatically

arXiv 2205.10364 ↗First posted 20 May 2022cs.LG cs.AI cs.CRcited by 11

Not on the map. Why?

Abstract (arXiv)

With the privatization deployment of DNNs on edge devices, the security of on-device DNNs has raised significant concern. To quantify the model leakage risk of on-device DNNs automatically, we propose NNReverse, the first learning-based method which can reverse DNNs from AI programs without domain knowledge. NNReverse trains a representation model to represent the semantics of binary code for DNN layers. By searching the most similar function in our database, NNReverse infers the layer type of a given function's binary code. To represent assembly instructions semantics precisely, NNReverse proposes a more fine-grained embedding model to represent the textual and structural-semantic of assembly functions.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 17 of its 20 references that are in the map. First posted on arXiv 20 May 2022.

Builds on

17 of 20 references.

References · 17

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  3. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  4. 2020Extracting Training Data from Large Language Modelscited by 2,144
  5. 2016Stealing Machine Learning Models via Prediction APIscited by 602
  6. 2018code2vec: Learning Distributed Representations of Codecited by 249
  7. 2019High Accuracy and High Fidelity Extraction of Neural Networkscited by 208
  8. 2018code2seq: Generating Sequences from Structured Representations of Codecited by 177
  9. 2019Thieves on Sesame Street! Model Extraction of BERT-based APIscited by 136
  10. 2018Glow: Graph Lowering Compiler Techniques for Neural Networkscited by 73
  11. 2018TVM: An Automated End-to-End Optimizing Compiler for Deep Learningcited by 62
  12. 2018Cache Telepathy: Leveraging Shared Resource Attacks to Learn DNN Architecturescited by 56
5 more
  1. 2020The Deep Learning Compiler: A Comprehensive Surveycited by 46
  2. 2018Stealing Neural Networks via Timing Side Channelscited by 42
  3. 2020Hermes Attack: Steal DNN Models with Lossless Inference Accuracycited by 41
  4. 2018Intel nGraph: An Intermediate Representation, Compiler, and Executor for Deep Learningcited by 22
  5. 2019Model Weight Theft With Just Noise Inputs: The Curious Case of the Petulant Attackercited by 8

Built on by

Cited by year

Citing papers per year: 2023: 7; 2024: 1; 2025: 3.

11 papers.

  1. 2023Sibling-Attack: Rethinking Transferable Adversarial Attacks against Face Recognitioncited by 32
  2. 2023Nova: Generative Language Models for Assembly Code with Hierarchical Attention and Contrastive Learningcited by 12
  3. 2023Dynamic Transformers Provide a False Sense of Efficiencycited by 11
  4. 2023DyCL: Dynamic Neural Network Compilation Via Program Rewriting and Graph Optimizationcited by 9
  5. 2024PPM: Automated Generation of Diverse Programming Problems for Benchmarking Code Generation Modelscited by 7
  6. 2025Your Compiler is Backdooring Your Model: Understanding and Exploiting Compilation Inconsistency Vulnerabilities in Deep Learning Compilerscited by 6
  7. 2023R^3: On-device Real-Time Deep Reinforcement Learning for Autonomous Roboticscited by 4
  8. 2023RED: A Systematic Real-Time Scheduling Approach for Robotic Environmental Dynamicscited by 4
  9. 2025THEMIS: Towards Practical Intellectual Property Protection for Post-Deployment On-Device Deep Learning Modelscited by 2
  10. 2025LeMix: Unified Scheduling for LLM Training and Inference on Multi-GPU Systemscited by 2
  11. 2023Dynamic Neural Network is All You Need: Understanding the Robustness of Dynamic Mechanisms in Neural Networkscited by 1