Joerg Appenzeller (left), Barry M. and Patricia L. Epstein Professor of Electrical and Computer Engineering, and Shreyas Sundaram (right), Marie Gordon Professor of Electrical and Computer Engineering

Two professors from Purdue University’s Elmore Family School of Electrical and Computer Engineering have been selected for Intel's 2023 Outstanding Researcher Awards (ORAs). Joerg Appenzeller and Shreyas Sundaram are among fifteen leading academic researchers to receive this honor. The annual award program recognizes exceptional contributions made through Intel university-sponsored research.

Appenzeller, Barry M. and Patricia L. Epstein Professor of Electrical and Computer Engineering, was chosen for his work on a project entitled “Performance Enhancers, Novel Device Structures and Fabrication Processes for Monolayer TMD p-FETs.” Appenzeller’s group is working on building high-performance p-channel metal oxide semiconductor (PMOS) tungsten-di-selenide transistors, a critical piece to realize a scaled front-end complementary metal oxide semiconductor (CMOS) transition metal dichalcogenide (TMD) technology. The 2D research community has struggled to achieve high currents in p-type transistors because of high contact resistance. Appenzeller’s team used hybrid charge-transfer and molecular doping to fabricate a transistor with some of the highest PMOS drive currents reported to date. They are also working on implementing a doped, raised source-drain structure, which has shown promising results.

Sundaram, Marie Gordon Professor of Electrical and Computer Engineering, is being recognized for his work on a project called “Robust and Scalable Runtime Optimization and Tuning for Diverse Computing Platforms.” Sundaram and his team addressed the problem of in-field adaptive control of large systems that scales to tens of control parameters in uncertain environments while meeting safety properties. The team developed advanced model-based and model-free approaches with a focus on sampling efficiency to enable fast convergence to improved configurations with reduced training data. A grey-box model-based reinforcement learning approach was developed that provides a rapprochement between standard model-based optimal control and modern data-driven learning. To address the cases where the target system model is not available, the team developed novel black-box reinforcement learning-based algorithms that balance exploration and exploitation with a focus on discrete action spaces to handle practical use cases.

Intel's ORAs are part of Intel's Corporate Research Council, which participates in research initiatives with prominent university science and technology centers and other ecosystem participants. Projects span the design of materials via machine learning, scalable optimization for diverse computing platforms, domain-specific compilers and novel accelerator architectures, metrology, field-programmable gate arrays (FPGAs), novel semiconductor devices, privacy-preserving knowledge transfer, and more. These Intel-sponsored research collaborations are advancing today's computing into future technologies.