Christopher M. Kim

This paper introduces DynML, a dynamical systems framework for modeling biological time series, showing that reservoir topological entropy correlates with predictive power.

The paper investigates how reversal probability is represented in prefrontal cortex neurons and recurrent neural networks during probabilistic reversal learning, finding stable initial states that drive non-stationary dynamics.

This paper presents optimized CPU and GPU implementations of the recursive least-squares algorithm for training spiking neural networks, enabling faster training of large-scale models.

The paper shows that task-related neural activity spreads across a cortical network through task-independent synapses, even in untrained neurons.

This study trains recurrent neural networks under strong coupling constraints to mimic cortical circuit dynamics, revealing biologically plausible connectivity and spiking behavior.

This study extends an EI neuron network model to include an E population with transmission delays, showing how critical coupling strength and delay modulate oscillatory and stationary states.

This study shows that modifying recurrent connectivity with a recursive least squares algorithm enables spiking networks to produce diverse spatiotemporal activity patterns, stabilize irregular spiking, and reproduce cortical neuron firing patterns during motor planning and movement.