Active learning of neural population dynamics using two-photon holographic optogenetics

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Advanced techniques like two-photon holographic optogenetics and calcium imaging allow precise neural population monitoring and perturbation.
• Studying neural population dynamics is crucial for understanding brain function.
• Current photostimulation experiments are time-consuming, with limited algorithmic approaches for optimizing stimulation patterns.

Purpose of the Study:

• To develop efficient methods for selecting optimal photostimulation patterns.
• To identify which neurons to stimulate to best inform dynamical models of neural activity.
• To reduce the amount of data required for accurate neural population modeling.

Main Methods:

• Development of an active learning procedure tailored for low-rank regression.
• Utilizing a low-rank linear dynamical systems model to analyze neural population responses.
• Application of the active stimulation design method to real and synthetic data from mouse motor cortex.

Main Results:

• Demonstrated the efficacy of the low-rank dynamical systems model for neural population activity.
• Showcased an active learning procedure that leverages low-rank structure to identify informative photostimulation patterns.
• Achieved up to a two-fold reduction in data required to reach a specific predictive power.

Conclusions:

• The developed active stimulation design method efficiently identifies informative photostimulation patterns.
• This approach significantly reduces the experimental data needed for modeling neural population dynamics.
• The underlying active learning procedure for low-rank regression may have broader applications.