講座摘要

院士講座系列 1

講者:生命科學組院士、國立清華大學生命科學院院長及腦科學研究中心主任
江安世院士
講題:Visualizing Memory Allocations in the Drosophila Brain

摘要:Memory formation requires allocation of learned information through new protein synthesis to specific neurons and synapses in a neurocircuit. To study the regulation of memory allocation, we need tools to visualize these sparse proteins in a large volume. Here, applying the improved Bessel beam plane illumination to the optically-cleared brain, we report a light-sheet super-resolution microscopy for single-molecule localization of learning-induced new proteins in the whole brain in Drosophila. We found that learning induces synthesis of vesicular monoamine transporter (VMAT) proteins in a single DPM neuron innervated all 15 lobe sectors of the mushroom body (MB), the memory center in the Drosophila brain. Within a single DPM neuron, these new VMAT molecules preferentially distributed throughout two out of the 15 MB sectors. Down-regulation of presynaptic VMAT or serotonin synthesis in the DPM neuron or postsynaptic serotonin receptors in Kenyon cells enhanced long-term memory. Hence, memory allocation outside the MB is mediated by serotonin release inside the specific MB sectors. This architecture supports spatiotemporal circuit coding of serotonin release in memory regulation. Our super-resolution imaging system has paved the way of studying memory allocation at synapse resolution in the whole brain.

院士講座系列 2

講者:工程科學組院士、國立交通大學材料科學與工程學系所特聘講座教授杜經寧院士

講題:Reliability challenges in mobile technology for big data era
摘要:In the era of big data and internet of things, the use of consumer electronic products in all aspects of our daily life is manifested by the ubiquitous presence of mobile devices.  The continuing demand for smaller size, more functionality, lower power consumption, and reduced cost is challenging.  A paradigm change, from 2-dimensional integrated circuits (2D IC) to 3D IC, is occurring in microelectronics industry.  The vertical interconnect in 3-dimension by using through-Si-via and micro-bump is a new packaging technique.  The design of the 3D IC mobile devices is in a fluid state because there is not enough field data on reliability to offer the optimal structure. The major challenge is whether we can use artificial intelligence (AI) to assist the finding of the best design with the required mean-time-to-failure.  In this talk, a feasible approach to address this challenge is discussed.