Codesign material, device, integrated circuits for bio-inspired, in-sensor processing

Departments: Electrical & Computer Engineering
Time: Wednesday, July 10, 2024 - 3:00pm - 4:00pm
Type:
Presenter: Dr. Houk Jang, Brookhaven National Laboratory
Room/Office: Becton 508A
Location:

Wednesday, July 10, 3:00pm ET

Becton 508A or Zoom: https://yale.zoom.us/j/98154437659

Hosted by: Fengnian Xia

Abstract: Natural intelligence (NI) develops through the continuous training of the brain, a vast natural neural network (NNN), by processing the incessant stream of information generated by numerous sensory organs. Similarly, artificial intelligence (AI), powered by software-based artificial neural networks (ANNs), follows this paradigm. The exponentially growing sensory networks, projected to reach 125 billion by 2030, form a crucial foundation for AI by generating and providing vast amounts of data. However, a significant portion of this data is unnecessary and redundant, causing substantial issues in data transmission and processing.

To overcome such bottlenecks, bio-inspired in-sensor processing has garnered significant research interests, as it generates and transmits processed data containing only essential information, similar to human sensory organs. In this seminar, I will discuss our codesign efforts spanning from materials and devices to integrated circuits for bio-inspired in-sensor processing. Specifically, I will cover the implementation of in-sensor processing capabilities enabled by integrating novel materials and devices, which have been excluded from traditional Si-based CMOS electronics, into large-scale integrated circuits. Examples include leveraging ionic memory effects in 2D materials and reconfigurable photodetectors, and their integrated circuits. These examples highlight the potential of harnessing exotic material/device properties at the circuit level to enable various in-sensor processing techniques.

Bio: Dr. Houk Jang is an assistant scientist at the Center for Functional Nanomaterials at Brookhaven National Laboratory, where he leads software development for QPress, an automated foundry for heterogeneous 2D material stacks. His research interests range from nanomaterials and novel optoelectronic/electronic devices to their integrated circuits for next-generation electronics, such as deformable electronics and bio-inspired computing. Houk’s pioneering contributions have garnered him a total citation count of over 18,000.