Biological connectomes as a representation for the Architecture of Artificial Neural Networks

@article{Schmidgall2022BiologicalCA,
  title={Biological connectomes as a representation for the Architecture of Artificial Neural Networks},
  author={Samuel Schmidgall and Catherine D. Schuman and Maryam Parsa},
  journal={bioRxiv},
  year={2022}
}
Grand efforts in neuroscience are working toward mapping the connectomes of many new species, including the near completion of the Drosophila melanogaster. It is important to ask whether these models could benefit artificial intelligence. In this work we ask two fundamental questions: (1) where and when biological connectomes can provide use in machine learning, (2) which design principles are necessary for extracting a good representation of the connectome. Toward this end, we translate the… 

References

SHOWING 1-10 OF 59 REFERENCES

Neural Circuit Architectural Priors for Embodied Control

This work translates C. elegans locomotion circuits into an ANN model controlling a simulated Swimmer agent and achieves good initial performance and asymptotic performance comparable with MLPs, while dramatically improving data quality and requiring orders of magnitude fewer parameters.

Connectome of the fly visual circuitry.

This paper reviews the recent study of connectomics in a brain of the fruit fly Drosophila and highlights how connectomes can provide a foundation for understanding the mechanism of neuronal functions by identifying the underlying neural circuits.

A critique of pure learning and what artificial neural networks can learn from animal brains

  • A. Zador
  • Biology, Computer Science
    Nature Communications
  • 2019
It is suggested that for AI to learn from animal brains, it is important to consider that animal behaviour results from brain connectivity specified in the genome through evolution, and not due to unique learning algorithms.

A connectivity model for the locomotor network of Caenorhabditis elegans

A new method for representing and analyzing the connectivity of a motoneuronal network was described, which led to a connectivity model for the neuromuscular network that generates locomotion in the nematode Caenorhabditis elegans.

Whole-animal connectomes of both Caenorhabditis elegans sexes

Quantitative connectivity matrices (or connectomes) for both adult sexes of the nematode Caenorhabditis elegans are presented that encompass all connections from sensory input to end-organ output across the entire animal.

SpikePropamine: Differentiable Plasticity in Spiking Neural Networks

The experimental results display that SNNs augmented with differentiable plasticity are sufficient for solving a set of challenging temporal learning tasks that a traditional SNN fails to solve, even in the presence of significant noise.

C. elegans locomotion: small circuits, complex functions

Meta-Learning through Hebbian Plasticity in Random Networks

This work proposes a search method that, instead of optimizing the weight parameters of neural networks directly, only searches for synapse-specific Hebbian learning rules that allow the network to continuously self-organize its weights during the lifetime of the agent.

A Perimotor Framework Reveals Functional Segmentation in the Motoneuronal Network Controlling Locomotion in Caenorhabditis elegans

The perimotor framework of observed connectivity and the segmented connectivity model give insights and advance the study of the neuronal infrastructure underlying locomotion in C. elegans and suggest that the tools used may be useful to interpret, simplify, and represent connectivity data of other motor systems.

Backpropamine: training self-modifying neural networks with differentiable neuromodulated plasticity

It is shown for the first time that artificial neural networks with neuromodulated plastic LSTMs with millions of parameters can be trained with gradient descent, and concludes that differentiable neurmodulation of plasticity offers a powerful new framework for training neural networks.
...