Yuqing Zhu

This paper compares gradient-based and evolutionary training methods in RNNs, showing they produce distinct connectivity and dynamics, especially in working memory tasks.

This study trains spiking neural networks to perform a state change detection task and finds that cross-stimulus-modulated inhibition emerges as an optimal solution under Dales law constraints.

The paper constructs spiking neural network models to show that cyclic transitions between triangle motifs are necessary for sustained activity in sparse, asynchronous networks.