–
December 16, 2019
Talk 1
Speaker
Yuri Rodrigues (MATHNEURO)
Title
A stochastic model of postsynaptic plasticity based on dendritic spine Ca2+ downstream proteins
Abstract
Dendritic spines are the hub in neuron-to-neuron excitatory communication and its efficiency can be changed by a timely organized spike sequence inducing plasticity changes in synapses. Preceding such changes a calcium influx is required to trigger a cascade of enzymatic events which ultimately will lead to an increase or decrease of synaptic efficiency. Still, it is not well understood how dendritic spines housing a noisy environment subject to stochasticity from multiple sources can translate a firing sequence into any specific transmission adjustments. Moreover, plasticity experimentation literature points out that not only the stochastic nature of plasticity gives an unclear comparison but also control parameters vastly differ among laboratories, for instance, temperature, brain region, the composition of artificial cerebrospinal fluid, specimen age. Trying to sort this out, we developed a mechanistic model to simulate the experimental conditions in which the calcium cascade in dendritic spines creates the required enzymatic pattern to induce plasticity observed in experiments.
Talk 2
Speaker
Yaohui Wang (STARS)
Title
Generative Adversarial Networks(GANs) for human video generation
Abstract
Generative Adversarial Networks (GANs) have been widely used in different tasks in computer vision, e.g., face image generation, super-resolution and colorization. However, generating realistic videos, especially videos related to human, have not been widely explored yet. In this talk, we will discuss our novel design of GANs to generate human videos, including facial expressions and human actions, which outperforms state-of-the-art methods based on all qualitative evaluation metrics and achieves the best visual quality.
