Functional Connectome: Approximating Brain Networks with Artificial Neural Networks
This work addresses a systems neuroscience problem by showing deep learning can model brain networks, though it appears incremental as it applies existing methods to new biological data.
The researchers tackled the problem of approximating biological neural circuit functions (functional connectome) using deep neural networks, achieving high accuracy in capturing computations from synthetic and empirically supported neural networks while demonstrating zero-shot generalization in novel environments.
We aimed to explore the capability of deep learning to approximate the function instantiated by biological neural circuits-the functional connectome. Using deep neural networks, we performed supervised learning with firing rate observations drawn from synthetically constructed neural circuits, as well as from an empirically supported Boundary Vector Cell-Place Cell network. The performance of trained networks was quantified using a range of criteria and tasks. Our results show that deep neural networks were able to capture the computations performed by synthetic biological networks with high accuracy, and were highly data efficient and robust to biological plasticity. We show that trained deep neural networks are able to perform zero-shot generalisation in novel environments, and allows for a wealth of tasks such as decoding the animal's location in space with high accuracy. Our study reveals a novel and promising direction in systems neuroscience, and can be expanded upon with a multitude of downstream applications, for example, goal-directed reinforcement learning.