However, with the expansion of the network, subject to the traditional Von Neumann architecture, the existing common hardware platforms (CPU, GPU and FPGA, etc.) It is hard to realize the related algorithms.Studies based on the new type of memristor have become the international hotspot and the subject frontier and can be widely used in the IoT and edge computing and intelligent recognition tasks.
Mesenchymal-based brain-like schematics.
Wu Huaqiang has long been devoting himself to relevant research and has profound technical accumulation.In the related report, Wu Huaqiang demonstrated how to optimize the joint design for brain-like computing based on devices, circuit structures and algorithms.According to material, device structure and process optimization , A memristor electronic synapse array with good bidirectional continuous resistivity change was prepared.Based on this array, experiments were performed to validate the face recognition function and to demonstrate the sustainable learning ability that is unique to other hardware implementations.
Wu Huaqiang further from the power consumption, area design, circuit design optimization, and for the characteristics of memristors, for the first time in the world proposed the hidden layer output binarization (Binarized-hidden-layer) chip architecture for handwritten digital body recognition Task, analyzed the influence of non-linearity, volatility and difference of the device, and put forward quantitatively the performance requirement to the memristor device, pointed out the direction of device optimization.
International Electron Devices Conference originated in 1955, the main coverage of international research and development in the field of microelectronic devices in the field of microelectronics is the top conference in the field of microelectronics, so far more than 60 years of history, in the international semiconductor technology enjoys a very high Academic status and broad influence, foreign media hailed as 'the field of micro-electronics Olympic gala'.