Kai Du

This paper demonstrates that synaptic credit assignment can be implemented online through voltage dynamics in detailed neurons, enabling supervised learning in recurrent networks.

The paper develops an analytical framework to understand how inhibition affects neuronal computation across dendritic compartments, revealing a unified law that explains inhibitory leverage in active dendrites.

MetaWorm integrates brain, body, and environment simulations to model C. elegans behavior, revealing how neural structure influences activity and movement.

This paper introduces a GPU-based framework that accelerates neuron simulations and applies it to neuroscience and AI tasks, showing significant speed improvements over traditional methods.

This paper introduces DeepDAM, a framework that improves neural connectivity inference by aligning synthetic and in vivo data domains, achieving high accuracy in rodent and other systems.

This paper introduces a computational framework to model synaptic adaptation in the early visual system, revealing diverse adaptive time constants and a frequency-sensitive gain-control strategy in ribbon synapses.

This paper develops a computational model of retinal gap junctions (G-filter) that improves deep neural networks' performance in blind denoising tasks.

The paper introduces U-RISC, a high-resolution EM dataset for cell membrane segmentation, and evaluates deep learning methods, finding a gap between current performance and human-level accuracy.

This paper introduces a general LSTM-based deep learning method, GNE, for automated parameter tuning of neuronal models and neural circuits, showing strong performance on both synthetic and experimental data.