一、交流项目招收对象
信电学院在读的优秀大二、大三年级学生，即2016、2017级本科生。
二、项目介绍
1. HKUST (Prof. Jiang Xu) 香港科技大学须江教授
Link: www.ece.ust.hk/~eexu
Project: Transfer Learning based Power Management among Multiple Devices
Opening: 2
Description:
Many human activities are fundamentally changed by high volumes of high-velocity high-variety information. Embedded systems in all forms are more popular now than ever before, and offer everyone the freedom to gather, access, store, process, and share information anytime and anywhere. Embedded systems, such as mobile phones and smart watches, offer a personal, portable, and intuitive form factor to access information freely. As application complexity and versatility burgeon, multiprocessor system-on-chip (MPSoC) has become the de facto platform for embedded systems, because integrating a system or a complex subsystem into a single chip improves performance and energy efficiency and lowers the cost per function.This project aims to further develop the transfer learning based power management among multiple embedded systems based on our previous work.
2. HKUST (Prof. Jiang Xu) 香港科技大学须江教授
Link: www.ece.ust.hk/~eexu
Project: Rejuvenate Post-Moore's Law Information Systems with Hybrid Electronics-Photonics
Opening: 2
Description:
Information systems, from supercomputer and data center to automobile, aircraft, and cellphone, are integrating growing numbers of processors, accelerators, memories, and peripherals to meet the burgeoning performance requirements of new applications under tight cost, energy, thermal, space, and weight constraints. Recent advances in hybrid electronics-photonics technologies promise ultra-high bandwidth, low latency, and great energy efficiency to alleviate the inter/intra-rack, inter/intra-board, and inter/intra-chip communication bottlenecks in information systems.Hybrid electronics-photonics technologies piggyback onto developed silicon fabrication processes to provide viable and cost-effective solutions. This project will systematically explore novel hybrid electronics-photonics technologies for post-Moore's Law information systems.
3. Nanostructured Gas Sensors
Abstract
Semiconductor nanostructures have interesting physical properties which makes them promising materials for novel electronics. Particularly, electronic devices can be potentially made smaller and consume less power with nanomaterials. In addition, nanostructures have much larger surface area than bulk materials, which makes them ideal candidates for gas/chemical sensing applications. This project aims to develop a novel nanostructure fabrication process, then utilize the semiconductor nanostructure for high performance sensors, which can be used for a wide range applications, for example, toxic gas detectors in smart home and smart city.
Tasks
- Learn nanostructure fabrication techniques
- Learn nanostructure device electrical characterization
- Learn gas sensor characterization
- Design sensor testing circuit and software
Deliverables
-Nanostructured gas sensors
- Sensor testing circuit and software
Supervisors:
Prof. Zhiyong Fan, eezfan@ust.hk. Room 2446, Tel: 2358-8027
Opening: 1
4. Optoelectronic Devices with Nanostructures
Abstract
Semiconductor nanostructures have unique physical properties which make them promising materials for novel electronics and optoelectronics. Particularly, optoelectronic devices such as photodetectors, LEDs, lasers can be potentially made smaller and consume less power with nanomaterials. This project aims to develop novel nanostructure fabrication process, then utilize the semiconductor nanomaterials for the above optoelectronic devices, which can be used for a wide range applications especially high performance and integrated optoelectronics.
Tasks
- Learn nanomaterial fabrication techniques
- Learn nanodevice fabrication techniques including microfabrication
- Learn device electrical and optical characterization
Deliverables
- Semiconductor nanostructure arrays
- Nanostructured light sensors and light emitting devices
Supervisors:
Zhiyong Fan, eezfan@ust.hk. Room 2446, Tel: 2358-8027
Opening: 1
5.
Contact:
Prof. Wei Zhang
Email: eeweiz@ust.hk
Webpage:http://www.ece.ust.hk/~eeweiz/
Efficient Machine Learning Design and Implementation on Accelerators
Opening: 1-2
Description:
Nowadays, Deep Neural Network (DNN) has attracted worldwide attention due to its great success in multiple areas ranging from object detection to speech recognition. However, the intensive computational complexity and high DDR bandwidth requirements of DNN have raised great challenges for its performance and power consumption both on the cloud and the edge. Hence, it is of great interests to accelerate DNNs on devices like GPUs, FPGAs, ASIC. In order to efficiently implement DNN on hardware accelerators, the DNN algorithms and its implementation need to be carefully designed considering the accelerator hardware characteristics and how to best utilize the hardware resources. The project will focus on the efficient DNN algorithm design and implementation techniques on the hardware accelerators considering low bit representation, sparsification, FFT transferring, etc.
6.
Contact:
Prof. Wei Zhang
Email: eeweiz@ust.hk
Webpage:http://www.ece.ust.hk/~eeweiz/
Point Cloud Processing on FPGA for Self-driving
Opening: 1-2
To achieve autonomous driving, a large amount of point cloud processing is performed for simultaneous localization and mapping (SLAM), action planning, object detection and tracking. Point cloud processing algorithms are computationally intensive and thus pose a challenge for unmanned platforms with limited space and energy. At present, microprocessor is usually used to execute the processing and several processors are needed to work together to accomplish the task, which is low efficiency in performance and power consumption. Use of field-programmable gate arrays (FPGAs) in point cloud processing is a promisingdirection. The project will focus on the development of FPGA based real-time point cloud processing system for the applications in autonomous vehicles.
选拔方式：
1. 申请者按项目的相关要求自己完成个人英文简历的设计，并将申请表（见附件）、中英文成绩单、语言证 明等相关材料电子版在3月15日前发钟老师邮箱：zhongtingting@zju.edu.cn；
（提示：所有资料合成一个pdf，文件命名为“年级+专业+姓名+导师+项目序号”）
2. 简历中必须明确说明个人已有的研究背景。
3. 教授将依据简历和相关材料进行遴选和面试。
五、说明
1．安全问题：入选者需要购买一份人身意外险，其他安全问题将由学生本人和家长自己承担，并向学院提供保险单复印件；
2. 申请人本人承担主要费用，包括往返机票，公寓租住及日常开销等。该项目将获得学校一定额度的资助。
如有疑问，可联系：钟老师87953027，QQ:408039838
申请表.xlsx