Developing co-located long-term and short-term memory for brain-inspired in-memory computing: a material-device-architecture co-design approach
| Interdisciplinary Areas: | Engineering-Medicine, Autonomous and Connected Systems |
|---|
Project Description
The uniqueness of the human brain in doing intelligent tasks at extremely low power comes from its ability to compute within the memory. One of the key challenges in mimicking the human brain for compute-in-memory comes from the lack of a co-located long- and short-term memory (LTM/STM), which enables reservoir computing (RC), a brain-inspired computing paradigm. The STM and LTM form core building blocks for the ‘reservoir’ and ‘readout’ layers respectively in a RC system. In this project, the post-doctoral scholar will be mentored to work across the material-device-architecture stack towards demonstrating the impact of RC paradigm on practical applications through experiments and device-to-system modeling. The project involves (1) optimizing material stack and developing a new biomimetic semiconductor device that exhibits reconfigurable co-located LTM and STM integrated into a novel three-terminal transistor; (2) building array-level prototypes as well as device-to-system modeling tools to demonstrate practical RC-based applications and to quantify the system-level benefits and tradeoffs The postdoc will collaborate with PhD students in adjacent projects to produce high-impact team research findings.
Start Date
01/13/2025
Postdoc Qualifications
1. PhD in Electrical Engineering/Materials Engineering/Computer Engineering with a research focus on semiconductor devices and/or systems.
2. Demonstrated ability to work on a multi-disciplinary project.
3. Semiconductor device fabrication/modeling/characterization experience is desired.
4. Knowledge and understanding of physical/architectural design of semiconductor chip is a plus.
5. Knowledge of AI/ML models is a plus.
Co-advisors
Raisul Islam
Assistant Professor of Materials Engineering
raisul@purdue.edu
https://engineering.purdue.edu/RISE-Lab
Haitong Li
Assistant Professor of Electrical and Computer Engineering
haitongli@purdue.edu
https://engineering.purdue.edu/NanoX/
Bibliography
1. DOI: 10.35848/1347-4065/ab8d4f
2. https://doi.org/10.1038/s41928-019-0313-3
3. https://doi.org/10.1002/adma.202204778
4. https://doi.org/10.1038/s41565-022-01095-3
5. https://doi.org/10.1007/s11467-023-1335-x