Intel® Rising Star Faculty Award 2022 Recognizes Research Efforts of 15 Leading Early-Career Professionals

Highlights

  • The Intel® Rising Star Faculty Award (RSA) program recognizes 15 early-career academic researchers leading groundbreaking technology research.

  • The program facilitates collaboration between award winners and leaders at Intel.

  • 2022 RSA winners are Adriana Schulz, Alessandro Lunghi, Chaya Ganesh, Christopher Brinton, Eilam Yalon, Lauren Garten, Mengjia Yan, Mingu Kang, Pratik Chaudhari, Rong Yang, Sophia Shao, Tianyin Xu, Virginia Smith, Xiang ‘Anthony’ Chen, Yiyang Li.

author-image

By

Each year, the Intel® Rising Star Faculty Award (RSA) program selects early-career academic researchers who lead groundbreaking technology research demonstrating the potential to disrupt the tech industry. The 15 award winners chosen for 2022 have presented impressive works in computer science, electrical engineering, computer engineering, material science, and chemical engineering.

The program proudly recognizes community members who are doing exceptional work in the field and hopes to facilitate long-term collaborative relationships with senior technical leaders at Intel. Award recipients are also chosen for their innovative teaching methods and for increasing the participation of women and underrepresented minorities in computer science and engineering.

The research conducted by the selected faculty provides novel solutions to challenges spanning various topics, including digital fabrication, quantum materials dynamics, cryptography, energy-efficient memory, microelectronics, computer architecture, deep learning, neuromorphic computing, and human-AI interaction.

This year’s winners include faculty members from the following institutions:

 

  • Carnegie Mellon University 
  • Cornell University
  • Georgia Institute of Technology
  • Indian Institute of Science, Bangalore
  • Massachusetts Institute of Technology 
  • Purdue University
  • Technion - Israel Institute of Technology
  • Trinity College Dublin
  • University of California, Berkeley
  • University of California, Los Angeles
  • University of California, San Diego
  • University of Illinois Urbana–Champaign
  • University of Michigan
  • University of Pennsylvania
  • University of Washington

Intel’s Rising Star Faculty Award Winners for 2022 (from top row, left): Christopher Brinton, Chaya Ganesh, Xiang ‘Anthony’ Chen, Sophia Shao, Pratik Chaudhari, Adriana Schulz, Mingu Kang, Lauren Garten, Alessandro Lunghi, Virginia Smith, Yiyang Li, Mengjia Yan, Tianyin Xu, Rong Yang, Eilam Yalon.

Christopher Brinton

Assistant Professor in the Elmore Family School of Electrical and Computer Engineering
Purdue University

Dr. Brinton’s research is at the intersection of distributed computing, wireless communications, network optimization, and machine learning. His work devises techniques for intelligence management in contemporary networked systems (i.e., networks for learning) and data-driven methodologies to optimize/defend how distributed systems operate (i.e., learning for networks). Over the next several years, he is planning to formalize “fog learning,” a new paradigm for training and managing machine learning models over contemporary fog network architectures. Unlike existing centralized and federated learning architectures, fog learning advocates intelligent orchestration of computing resources across network elements spanning the “cloud-to-things continuum” from datacenter servers to edge devices. The improvements in model quality, resource efficiency, and network security provided by fog learning will help enable widespread deployment of our world’s increasingly complex deep learning models across mobile devices, thus revolutionizing edge intelligence and mobile user quality of experience. Brinton is also the recipient of the National Science Foundation’s CAREER Award, the Office of Naval Research’s YIP Award, and the Defense Advanced Research Projects Agency’s YFA Award.

Chaya Ganesh

Assistant Professor in the Department of Computer Science and Automation
Indian Institute of Science, Bangalore

Ganesh hopes to design new constructions for zero-knowledge proofs that are more suitable for blockchain and other decentralized applications. She envisions a stateless blockchain powered by authenticated data structures providing verifiable query access where the state is distributed among all users. As cryptography moves from theory to practice, Ganesh has identified the analysis of cryptographic primitives deployed in practice as an essential area of research, with a larger goal to study the extent and limits of subversion of various primitives and protocols. In addition, the adoption of machine learning in many applications today, such as face recognition, threat prediction, credit rating, etc. has led to a shift in societal power. This shift raises questions about the fairness and verifiability of classification decisions, security of a model, and privacy of data used to train a model. Over the next five years, Ganesh aims to provide systematic answers to such emerging questions in machine learning through cryptography.

Xiang ‘Anthony’ Chen

Assistant Professor in the Electrical and Computer Engineering Department 
University of California, Los Angeles

The development of intelligence is built upon the evolution and passage of human knowledge over generations. Unfortunately, the current training of AI is largely deprived of such human knowledge. Humans often provide only a single label for each training data point and expect AI to ‘reverse-engineer’ the underlying human knowledge, at times causing biases and errors when such knowledge is misrepresented. Xiang Chen’s hopes to expand the interaction bandwidth between human and AI with three thrusts. First, making AI comprehensible to humans, then enabling humans to control AI (e.g., by conveying knowledge to AI),  and finally, by supporting collaboration between humans and AI. In the next five years, Chen hopes to accomplish this goal by focusing on application domains where a human knowledge worker needs to distill insights by interacting with a large data view, e.g., physicians examining medical images, biologists discovering wildlife from video footage, and astronomers studying planetary behavior from telescope data.

Sophia Shao

Assistant Professor in the Electrical Engineering & Computer Sciences Department
University of California, Berkeley

Sophia Shao's research centers around improving the scalability, efficiency, and programmability of heterogeneous platforms from edge devices to data centers. Her past work has led to the development of the first multi-chip-module-based, deep-learning inference accelerator that delivers impressive performance and energy efficiency, and novel program analysis methodologies to help designers navigate the large design space of specialized, heterogeneous systems. Shao is building on her earlier efforts to tackle the next unsolved frontiers in specialized hardware: how to effectively deploy a large number of heterogeneous accelerators at the system level. Shao’s recent work has led to the development of Gemmini, a modular accelerator generator of spatial-array-based architectures, and CoSA, a constraint-optimization-based approach to program machine-learning accelerators. Moving forward, Shao will explore the intersection of architectural prototyping, algorithm development, and programming support for heterogeneous accelerators.

Pratik Chaudhari

Assistant Professor of Electrical and Systems Engineering
University of Pennsylvania

Pratik Chaudhari’s goal is to understand learning in artificial and biological systems by identifying common themes in intracellular signaling networks of proteins, retinal circuits, and artificial neural networks. He proposes to investigate a characteristic structure called “sloppiness”. Sloppy systems are neither completely robust (insensitive to perturbations but difficult to adapt), nor perfectly tuned (efficient and precise but fragile). Instead, they exhibit a wide range of sensitivities. His work aims to show how sloppiness allows a system to strike a balance between doing well on one task and adapting to new ones. To accomplish this, Chaudhari proposes research organized into two main thrusts. First, examining how sloppiness can explain the performance of artificial neural networks, which could lead to a transformative understanding of the conceptual basis of deep learning. Second, studying models of the early visual system to understand how sloppiness emerges when circuits of neurons adapt to changes in the environment; this can inform the design of algorithms for neuromorphic hardware such as event-based cameras.

Adriana Schulz

Assistant Professor in the Department of Electrical and Computer Engineering
University of Washington

Adriana Schulz invents algorithms that address fundamental technical challenges in computational design and uses them to build end-to-end design systems to help drive the manufacturing revolution. These solutions not only increase productivity but dramatically improve the quality of the products themselves. Schulz’s next-generation design tools for manufacturing intelligently guide users through the design process, enable users to optimize how the objects they are creating will behave in the physical world, and help users integrate the manufacturing process itself into the design process. Thus, fundamentally changing what can be made and by whom. She plans to further disrupt design software by developing novel approaches that combine numeral optimization, machine learning, and programming languages. However, recognizing that the current manufacturing infrastructure causes significant damage to our environment and natural resources, Schulz also hopes to revolutionize manufacturing using computational design algorithms that can guide product designers to reduce waste at the source.

Mingu Kang

Assistant Professor in the Department of Electrical and Computing Engineering 
University of California, San Diego

The primary objective of Mingu Kang’s lab is to research energy- and latency-efficient integrated circuits, architectures, and systems for machine learning and various signal processing algorithms by leveraging non-von Neumann approaches including in-memory computing, in-sensor computing, and neuromorphic computing with both CMOS and emerging devices. His group pursues vertically integrated cross-layered research methods by focusing on a VLSI circuit and micro-architecture design at its center, and connecting other layers such as device, algorithm, and large systems. His group also emphasizes thorough validation of the research ideas via prototyping in silicon to deliver more accurate and trustworthy scientific findings. Kang’s group has a grand research vision for the next following years to enable the heterogenous integration of analog and digital computation for an ultralow-power AI processing through his vertically integrated research efforts.

Lauren Garten

Assistant Professor in the School of Materials Science and Engineering
Georgia Institute of Technology

Dr. Lauren Garten is an assistant professor in the School of Materials Science and Engineering at the Georgia Institute of Technology.Employing novel processing methods and unique characterization techniques, Lauren Garten investigates how the cross-coupling of materials properties can enhance multiferroic device performance. She has previously discovered clean-room-accessible, CMOS compatible ferroelectrics, and is currently developing 2D ferroelectrics and multiferroics. These materials create new routes to enhance and control charge transport based on ferroelectric polarization-charge coupling. The next step is translating these materials’ properties into new device architectures. Over the next five years, Dr. Garten plans to develop world-class programs at the nexus of multiferroics, ferroelectric, and electromechanical materials and devices.

Alessandro Lunghi

Assistant Professor in the School of Physics 
Trinity College, Dublin 

Having pioneered the field of ab-initio spin dynamics simulations for magnetic semiconductors, Alessandro Lunghi is now spearheading the development of machine-learning strategies for designing new magnetic materials for quantum technology applications. In particular, Lunghi has devoted significant efforts to studying magnetic molecules as potential elements of quantum devices and classical high-density magnetic memories. During his Ph.D., he has been the main driver of developing a computational ab-initio strategy for the prediction of spin-phonon relaxation. From explaining the physics of spin relaxation to providing new guidelines for synthesizing new compounds, Lunghi’s results have effectively changed the conceptualization of spin relaxation in localized magnetic systems. His plans for the future aim at revolutionizing current design strategies for magnetic molecules with application in the space of quantum information science. Lunghi is planning to use machine learning and high-throughput simulations to efficiently explore the chemical space of coordination compounds in search of materials with outstanding quantum coherence.

Virginia Smith

Assistant Professor in the Machine Learning Department and an affiliated faculty in the Department of Electrical & Computer Engineering
Carnegie Mellon University

Mobile phones, wearable devices, and smart homes form just a few of the modern distributed networks generating a wealth of data each day. Due to the growing computational power of edge devices, coupled with concerns over transmitting private data, it is increasingly attractive to store data locally and push network computation to the edge. Federated learning stands to disrupt the computing industry by allowing for a wealth of knowledge to be shared across remote sources, but only if this information can be shared while maintaining user privacy, ensuring quality of service, and adhering to strict computational constraints. Virginia Smith’s research aims to provide such guarantees for federated learning, ultimately making the process of learning over a massive federated network as seamless and secure as learning over centralized data. Smith has spearheaded some of the first works analyzing the effects of heterogeneity in distributed learning; co-developed LEAF (leaf.cmu.edu), a popular benchmarking framework for learning in federated settings; and co-organizes a weekly worldwide federated learning seminar (FLOW).

Yiyang Li

Assistant Professor of Materials Science and Engineering
University of Michigan

Yiyang Li is building a materials research program that can be fundamentally disruptive to semiconductors by incorporating electrochemical principles. A grand challenge for brain-inspired neuromorphic computing that can improve AI energy efficiency is to develop an analog nonvolatile memory. Li is investigating how ions move to store information in memristors, a promising solution for analog memory. He was also one of the inventors of the electrochemical random access memory (ECRAM). Like a lithium-ion battery, ECRAM operates through ionic motion in solids. Li recognized that the state of charge of a battery is an analog value (100%, 99%, etc); ECRAM utilizes this intrinsically analog nature of ionic materials to store analog memory in a solid-state device. Li has also applied microelectronic principles to batteries. Device-to-device variability is an important metric in microelectronics. However, it has not been possible to directly quantify variability in batteries. Motivated by the need to measure “particle-to-particle variability,” Li is building novel instrumentation platforms able to study the electrochemical variability by cycling one battery particle at a time.

Mengjia Yan

Assistant Professor of Electrical Engineering and Computer Science
Massachusetts Institute of Technology

Security vulnerabilities that breach processor security and broadly affect computer systems have become one of the central security threats in computer architecture. Many of these attacks employ speculative execution to bypass memory isolation and leak arbitrary data. Unlike other attacks that can be addressed with simple patches, speculation-based attacks exploit fundamental design principles of hardware processors making nearly all modern processors vulnerable to these attacks. Mengjia Yan is at the forefront of efforts to address these, and related vulnerabilities. Her research vision is to rethink computer architecture from the ground-up for security. She has proposed InvisiSpec, the first comprehensive hardware defense solution against speculative execution attacks in multiprocessor cache hierarchies. Yan has already applied her framework successfully to caches, and next she plans to extend it to other side-channels such as branch predictors, TLBs and several other structures. Her future work in this area will focus on generalizing both the technique and analysis to comprehensively understand the tradeoffs between performance and amount of information that can be leaked.

Tianyin Xu

Assistant Professor in the Engineering and Computer Science Department
University of Illinois Urbana – Champaign

Given the high rate of sharing in cloud and datacenter computing, Tianyin Xu recognizes that security is a pressing concern for cloud and datacenter computing, and has centered his recent work on operating system (OS) security. Addressing a major barrier for implementing fine-grained security policies, he and his collaborators developed a novel cache system to minimize performance overhead and enable comprehensive security checking. Xu continues to lead the research on system-call security with the focus on building more powerful and more efficient security. He has started an ambitious project on revamping the OS design and implementation for extensible and customizable virtual memory by building extensible, programmable interface and modularized construction. Over the next five years, his project aims at making the virtual memory subsystem easy to customize and extend to accommodate the emerging needs. Xu also plans to design new virtualization technologies at the layers of OS and cluster management systems to effectively and securely support heterogenous and hybrid cloud.

Rong Yang

Assistant Professor in the Robert Frederick Smith School of Chemical and Biomeolecular Engineering
Cornell University

Poised to disrupt the computing industry, Rong Yang’s research has transformed the way we think about chemical vapor deposition (CVD)-based manufacturing of functional soft materials.
Yang’s research has enabled dozens of novel dielectric and/or ion-conducting polymer thin films, obtained using her all-dry synthesis technique at high purity, ultra-fast deposition rate, and exceptional conformality over nano/microstructures. Her research goal in the next five years is to advance neuromorphic computing via CVD-based materials design, synthesis, and device fabrication. To meet the increasingly demanding computational density and energy requirements of artificial neural networks, she proposes to enable novel electrochemical random-access memories (ECRAMs) using the polymer thin films synthesized using her CVD polymerization techniques. Her proposed research could transform the manufacturing science for neuromorphic computing devices by enabling a one-step and bottom-up synthesis platform for the channel, gate, and solid electrolyte materials in a ECRAMs, which delivers rapid deposition (~1-100 nm/min) with nanoscale control of the coating thickness while eliminating solvents and impurities.

Eilam Yalon

Assistant Professor in the Faculty of Electrical and Computer Engineering
Technion, Israel Institute of Technology

Eilam Yalon is currently engaged in research regarding nano-electronics, semiconductor device physics, energy-efficient electronics, emerging non-volatile memory technology, power and heat dissipation in devices, two-dimensional (2D) materials, and their application in electronic devices. The long-term goal of his research is to address technological challenges in nanoscale devices by better understanding current and heat flow in nanomaterials and their interfaces. The work focuses on materials that do not require epitaxial growth and can be monolithically stacked and integrated. Specific materials are known to be advantageous for particular application domains. For example, 2D materials are suitable for ultra-scaled energy-efficient transistors, phase change, and ferroelectric materials are ideal for fast, low-power memories, etc. Still, their integration is crucial and can enable novel devices and functionalities. Yalon’s near-future research agenda consists of natural extensions of the ongoing studies as well as new research directions. His team will use the platform developed for nanoscale thermometry to study the dynamics of phase change memory and threshold switching materials of interest. This combination of materials and device physics study will help identify the next-generation devices for emerging applications beyond non-volatile memory, such as neuromorphic hardware.