Neural Network Quine
Open accessThe DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index.
Abstract
Self-replication is a key aspect of biological life that has been largely overlooked in Artificial Intelligence systems. Here we describe how to build and train self-replicating neural networks. The network replicates itself by learning to output its own weights. The network is designed using a loss function that can be optimized with either gradient-based or non-gradient-based methods. We also describe a method we call regeneration to train the network without explicit optimization, by injecting the network with predictions of its own parameters. The best solution for a self-replicating network was found by alternating between regeneration and optimization steps. Finally, we describe a design for a self-replicating neural network that can solve an auxiliary task such as MNIST image classification. We observe that there is a trade-off between the network's ability to classify images and its ability to replicate, but training is biased towards increasing its specialization at image classification at the expense of replication. This is analogous to the trade-off between reproduction and other tasks observed in nature. We suggest that a self-replication mechanism for artificial intelligence is useful because it introduces the possibility of continual improvement through natural selection.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“While a quasi-reflexive framework implemented via neural Quines [72] or Hyper networks [73] can manifest such approximate topologies”
From Self-Referential Introspection in Large Language Models: The Critical Threshold for Recursive Self-Improvement (2026) · cites this paper as [72]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 4×
“We trained a vanilla quine with classical SGD (lr = 0.01), SGD with momentum (lr = 0.01, ρ = 0.9), ADAM (Kingma and Ba, 2014), Adagrad (Duchi et al., 2011), Adamax (Kingma and Ba, 2014), and RMSprop (Tieleman and Hinton, 2012) with default hyperparameter settings on the self-replicating loss for 30 epochs.”
BackgroundMethodInfluentialDelving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.123 · Cited by 19,741 · cited 1×
“The quine was initialized with the same procedure as in He et al. (2015), and the initial loss LSR prior to any training was 90.16.”
MethodInfluentialExtreme learning machine: Theory and applications
Huang, G.-B., Zhu, Q.-Y., & Siew, C.-K. (2006) · Neurocomputing · doi:10.1016/j.neucom.2005.12.126 · Cited by 13,285 · cited 1×
“Indeed, they form a key component of Extreme Learning Machines (Huang et al., 2006) which are feedforward neural networks that have proven useful in classification and regression problems.”
BackgroundUnsupervised representation learning with deep convolutional generative adversarial networks
Radford, A., Metz, L., & Chintala, S. (2015) · arXiv · doi:10.48550/arxiv.1511.06434 · Cited by 12,235 · cited 2×
“We also looked at open-source implementations of two popular generative models for images, DCGAN (Radford et al., 2016) and DRAW (Gregor et al.”
Method(2010) · Cited by 8,261
Gulshan, V., Peng, L., Coram, M., Stumpe, M. C., Wu, D., Narayanaswamy, A., Venugopalan, S., Widner, K., Madams, T., Cuadros, J., et al. (2016) · JAMA · doi:10.1001/jama.2016.17216 · Cited by 7,697
Approximation capabilities of multilayer feedforward networks
Hornik, K. (1991) · Neural Networks · doi:10.1016/0893-6080(91)90009-t · Cited by 6,091 · cited 1×
“MLPs are known to be good function approximators, specifically a feedforward neural network with at least one hidden layer forms a class of functions that is dense in the space of continuous functions under a compact domain (Hornik, 1991; Cybenko, 1989).”
BackgroundWaveNet: a generative model for raw audio
Oord, A. V. D., Dieleman, S., Zen, H., Simonyan, K., Vinyals, O., Graves, A., Kalchbrenner, N., Senior, A., & Kavukcuoglu, K. (2016) · arXiv · doi:10.48550/arxiv.1609.03499 · Cited by 4,911 · cited 1×
“, 2016), synthesizing human-like speech (Shen et al., 2017; Oord et al., 2016), and executing strate-”
BackgroundOptimization as a Model for Few-Shot Learning
Ravi, S., & Larochelle, H. (2017) · International Conference on Learning Representations · Cited by 4,067
Natural TTS Synthesis by Conditioning Wavenet on MEL Spectrogram Predictions
Shen, J., Pang, R., Weiss, R. J., Schuster, M., Jaitly, N., Yang, Z., Chen, Z., Zhang, Y., Wang, Y., Skerrv-Ryan, R., et al. (2018) · 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2018.8461368 · Cited by 2,791 · cited 2×
“, 2016), synthesizing human-like speech (Shen et al., 2017; Oord et al., 2016), and executing strate-”
Background
Research works printed but not linked: 8
- Bengio, Y., Courville, A., and Vincent, P. (2013). Representation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence , 35, no. 8:17981828.
- Cybenko, G. (1989). Approximation by superpositions of a sigmoidal function. Mathematics of control, signals and systems , 2(4):303-314.
- Ha, D., Dai, A. M., and Le, Q. V. (2017). Hypernetworks. International Conference on Learning Representations .
- Schmidhuber, J. (1993). A self-referentialweight matrix. In ICANN93 , pages 446-450. Springer.
- Stanley, K. O., DAmbrosio, D., and Gauci, J. (2009). A hypercubebased indirect encoding for evolving large-scale neural networks. Artificial Life , 15(2):185-212.
- Tieleman, T. and Hinton, G. (2012). Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude. COURSERA: Neural networks for machine learning , 4(2):26-31.
- Von Neumann, J. and Burks, A. W. (1966). Theory of selfreproducing automata. page 8. Urbana: University of Illinois Press.
- Wang, T., Sha, R., Dreyfus, R., Leunissen, M. E., Maass, C., Pine, D. J., Chaikin, P. M., and Seeman, N. C. (2011). Selfreplication of information-bearing nanoscale patterns. Nature , 478(7368):225.
Cited by
Williams, M., & Braddock, M. (2019) · Studia Humana · doi:10.2478/sh-2019-0001 · Cited by 77
MOSFET Physics-Based Compact Model Mass-Produced: An Artificial Neural Network Approach
Huang, S., & Wang, L. (2023) · Micromachines · doi:10.3390/mi14020386 · Cited by 28
Humans, super humans, and super humanoids: debating Stephen Hawking’s doomsday AI forecast
Kumar, S., & Choudhury, S. (2022) · AI and Ethics · doi:10.1007/s43681-022-00213-0 · Cited by 25
“Their proposed network learns to output its weights and so replicates itself [10].”
BackgroundLifelong learning in artificial neural networks
Anthes, G. (2019) · Communications of the ACM · doi:10.1145/3323685 · Cited by 16
Winkler, E. A. (2022) · SN Social Sciences · doi:10.1007/s43545-022-00350-7 · Cited by 21
“2018; Chang and Lipson 2018; Copeland 2021) and artificial consciousness (Graziano 2013; Haikonen 2019); the dehumanization of human enhancement technologies (transhumanism) (Fukuyama 2004; Miah 2012; Savulescu and Bostrom 2009); the abolition of human dignity; systemic loss, violation or…”
Background
Metadata
- Authors
- Chang, O., & Lipson, H.
- Affiliations
- Columbia University (US)
- Venue
- The 2018 Conference on Artificial Life
- Publisher
- MIT Press
- Published
- 2018
- Pages
- 234-241
- OpenAlex
- W2792712144
- Cited by
- 16
- References
- 28
- Citations by year
- 2026: 2 · 2023: 1 · 2022: 7 · 2021: 1 · 2020: 1 · 2019: 3 · 2018: 1
- FWCI
- 0.61
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Conference on Artificial Life (ALIFE) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Evolutionary Algorithms and Applications
- Physical Sciences › Computer Science › Computational Theory and Mathematics › Cellular Automata and Applications
- Keywords
- self-replication, neural network weights, regeneration, loss function design, gradient-based optimization, artificial life
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Neural Network Quine. https://app.cochat.ai/papers/10.1162/isal_a_00049
- The paper
- Chang, O., & Lipson, H. (2018). Neural Network Quine. The 2018 Conference on Artificial Life. https://doi.org/10.1162/isal_a_00049
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.