The Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship, a program of Schmidt Futures, is a new international initiative that seeks to change how science is done by accelerating the incorporation of AI techniques into the natural sciences, engineering, and mathematical science (STEM), providing access to AI tools and training to the sharpest minds on the frontlines of scientific innovation.
The program aims to create great science; an acceleration of the substantive use of AI in imagining, planning, executing, and supporting scientific research around the world; and the building of a global cohort of scientific leaders with a grounding in the application of AI in their fields.
Fellows will have the opportunity to pursue original research on significant questions in AI and science. They will form a cohort of top scholars across the natural sciences engaging in joint training and research activities. Fellows will be provided with a unique, cross-campus ecosystem of training in AI methods necessary for conducting their research. They will receive a competitive salary and benefits, generous travel allowances, and the opportunity to collaborate with partners worldwide.
- Applicants must be within 3 years of having obtained PhD degree or equivalent degree in non-computer science areas.
- Excellent academic records and research potential.
- Applicant is required to submit a thoughtful and realistic inter-disciplinary research proposal. Only research on applying AI techniques to engineering, natural sciences and mathematical sciences will be supported.
- Applicants are required to identify a potential project mentor (i.e. NTU Faculty), who can provide access to valuable academic and career guidance.
Fellowship Terms and Conditions
- The fellowship will be tenable for up to two (2) years with an attractive remuneration package.
- Awardees will be required to take at least one AI module every year during the fellowship.
- Awardees are required to submit an annual progress report to the University and Schmidt Futures, stating their progress, results and contributions.
- Awardees are required to attend conferences/workshops/seminars organized by the host University and Schmidt Futures to present their research and seek collaborations.
- The Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship 2023 Call for Application is currently closed.
- Candidates interested to apply for the scheme are required to contact their potential project mentor (i.e. NTU Faculty) and work with the potential mentor in submitting their application package. Please check with the project mentor for internal deadline.
- Each applicant is required to submit the complete application form (Download Template) with the following supporting documents:
- 2-page research proposal (Section 7 (i) on application form)
- Comprehensive CV (with full publication list)
- Official Doctorate transcript or Degree Certificate in English
- Confirmation from NTU faculty member who is agreeable to be the applicant’s project mentor.
- Two reference letters (one must be from PhD Supervisor)
- Top 3 publications (in PDF) and any other supporting documents (scanned copy)
Queries can be sent to Dr Yao Wei ([email protected])
Host School : SPMS
Email: [email protected]
Google Scholar: https://scholar.google.com/citations?user=D_YienkAAAAJ
Yang Long obtained his Ph.D. in Physics from Tongji University, Shanghai, China in 2020. Currently, he is a research fellow at Nanyang Technological University, Singapore. Dr. Long's research expertise includes investigating spin angular momentum and
new topological phenomena in classical wave systems, directional control of near-field photonics/phononics, and machine learning in fundamental physics. He is currently concentrating on the classification of topological phases of matter without human
knowledge and developing new kinds of unsupervised learning algorithms for replacing complicated abstract mathematics.
- Machine learning in physics
- Topological physics
- Near-field wave physics
- Spin angular momentum in classical waves
Project: Machine learning of topological classifications without human knowledge
Abstract: A significant and unfading theme in fundamental science is the classification of matter. The classifications of matter reflect the uses and functions of matter. However, the traditional classifications of matter are primarily restricted to using a few common physical characteristics or degrees of freedom, such as elements, geometrical structures, or chemical compounds. Topological classification, a new classification scheme based on topology, has been gradually introduced into investigations of classifications of matter. Natural matter can be further divided into various topological classes, such as topological/trivial insulators and topological semimetals, in accordance with topological classifications. Novel topological phenomena between two topologically distinct materials have been observed with plenty of applications such as topological laser and on-chip THz-region optical transport. Although many theories about topological classifications have been constructed, there are still new topological mechanisms or materials reported very recently. The reason is that these theoretical approaches heavily rely on abstract mathematics (such as homotopy group, K-theory, or Clifford algebra), which is incomplete, under slow development, and may be incorrect in new cases. Note that many materials previously classified as “trivial” are found to be topological with new theoretical advances. Therefore, a question is raised: is it possible to dispense with flawed mathematical techniques (e.g., homotopy group and K-theory) or refrain from using any human knowledge? In other words, can we realize topological classifications only based on raw data collected from matter? In this project, I aim to systematically incorporate machine learning to realize topological classification in a data-driven manner, in order to abandon the traditional abstract mathematics. The methodology is based on applying unsupervised learning technologies to raw data from Hamiltonian samples to capture topologically distinct phases. In contrast to earlier math-heavy theories, this data-driven approach will not require any human experience or knowledge. Therefore, it will not miss a “hidden” topological phase due to the incomplete list of topological invariants or theoretical limitations. It also takes practical constraints (e.g., a finite number of bands) into account, and reveals some previously unnoticed features.
Dao Fuying received a Ph.D in Biomedical Engineering from the University of Electronic Science and Technology of China. Based on a background in biology and computational science, Fuying’s doctoral research focused on origin replication sites identification
in eukaryotic genome through integrative analyses of genome-wide DNA methylation, histone modification profiles, and 3D genomics (Dao et al. 2023, INT J BIOL MACROMOL;
Dao et al. 2022, RESEARCH; Dao et al. 2021a,BRIEF BIOINFORM
Highly cited paper); Dao et al. 2021b,BRIEF BIOINFORM (Highly cited paper); Dao et al. 2021c,BRIEF BIOINFORM;
Dao et al. 2019, Bioinformatics (Highly cited and hot paper)). Fuying continued her research
in Dr. Melissa fullwood’s lab as a postdoctoral research fellow, to identify chromatin interactions and cancer biomarkers using artificial intelligence methods.
Research Interests: Dao Fuying is interested in designing and using computational approaches to solve biological issues. Particularly, she prefers to identify chromatin interactions and cancer biomarkers using artificial intelligence methods, aim to develop precise therapies for cancer by identifying specific chromatin interaction-based biomarkers.
Project: Application of deep learning in identifying chromatin interactions and cancer biomarkers
Abstract: In eukaryotes, chromatin is folded into complex 3D structures and dynamically regulates life processes. Use of artificial intelligence methods to identify specific chromatin interaction-based biomarkers enables the differentiation of various cancer subtypes and the development of precise therapies for more effective cancer treatment. Here I hypothesize that a deep learning model can mine informative features and predict accurately chromatin interactions in RNA-Seq samples. The significance of this work is that if successful, we will be able to analyze large cohorts of clinical RNA-Seq data for chromatin interactions and suggest potential epigenetic drugs for further drug testing.
Host School : CEE
Email: [email protected].
Tian Qingyun obtained her B.Eng degree from Southeast University, China in 2016. Then she received her Ph.D. degree from Nanyang Technological University, Singapore, in 2022. After graduation, she worked as a research fellow in Nanyang Technological University and National University of Singapore. Her research mainly focuses on modelling and optimization of transportation system.
Research Interests: Urban transportation, Connected and autonomous vehicles, Infrastructure planning and design, Shared mobility, Intelligent transportation system.
Project: Learning based modeling framework and application in operation management of transportation system
Abstract: In presence of a rapidly increasing urban population and the resultant travel demand expansion, many major cities in the world are facing the urban challenge of offering efficient mobility services for both passengers and goods. Nowadays, with the development of technology and high-performance computing, the smart transportation system is being popular in both academia and practice. AI technology based on "learning" from massive amounts of data can identify features and make analysis and predictions more quickly, which provides new ways for transportation planning, operation, and management. The research project focuses on the application of AI in transportation. The objective is to provide solutions to the operation management of transportation systems. To achieve the goal, AI technology, machine learning, and transportation model will be required to be integrated into the whole framework.
Florian Rossmannek obtained his bachelor's, master's and PhD degree in mathematics from the Swiss Federal Institute of Technology in Zurich in 2018, 2019, and 2023, respectively. During his doctoral studies he was a member of RiskLab Switzerland. Now he is a Schmidt AI in Science Postdoctoral Fellow at Nanyang Technological University Singapore.
Research Interests: approximation and optimization problems in machine learning, neural networks, reservoir computing, applications of machine learning to quantum physics.
Project: Learning Schrödinger’s equation with reservoir systems
Abstract: Machine learning has seen many applications in which it served as a black box to generate good results. To improve upon this and make the black box interpretable, one needs to understand the application of interest mathematically and tailor the machine learning approach accordingly. In this project, we pursue this methodology for applications in quantum physics. One difficult challenge therein is to find the so-called ground and excited states, which are energy levels of the system. Knowing these states is fundamental to understanding and being able to simulate chemical reactions. Mathematically speaking, we design a new interpretable machine learning approach based on reservoir computing to learn eigenvalues of a Hamiltonian operator by considering Schrödinger's equation.
Wadgaonkar Indrajit Pradeepchandra
Host School : SPMS
Email: [email protected]
Indrajit received a PhD in Physics from the Nanyang Technological University (NTU), Singapore in March 2023. During his PhD, he worked on the application of the semi-classical framework (Time Dependent Boltzmann Equation) to decipher ultrafast dynamics
in quantum materials. Before joining NTU, Indrajit completed his master's at the Indian Institute of Technology, Madras, India and then worked as a senior manager for research and development with India's largest automobile manufacturer, Tata Motors
Ltd. During his stint at Tata Motors Ltd. Indrajit spearheaded the computational fluid dynamics analysis at the Advanced Engineering department wherein he worked primarily on fuel-cell technology for the future. Indrajit is a recipient of the Government
of India Scholarship for higher studies, the J.N.Tata Endowment Scholarship (2018), Nanyang Research Scholarship (2018), the first Imperial-TUM-NTU Global Fellows Programme(2020) and The Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship
Ultrafast dynamics in quantum materials, semiclassical analysis of quantum systems, theoretical quantum physics and statistical mechanics
Project: Artificial Intelligence assisted state of the art modelling of phonon Poiseuille flow in graphene and thin graphite
Abstract: Clock speed and hence processor speed for modern electronics has been stagnant for the last two decades owing to restrictions on the maximum rate of heat dissipation, which in turn is a result of the limited thermal conductivity of component materials. In principle, hydrodynamic Poiseuille flow of phonons in graphene promises an infinite thermal conductivity which can be potentially engineered to realise revolutionary technological applications in modern electronics. However, the inherent complexity in describing this phenomenon has so-far thwarted the development of real-world applications by preventing a satisfactory numerical model which can augment the on-going experimental advances. Although the Time Dependent Boltzmann Equation (TDBE) is a widely acknowledged framework to model hydrodynamic transport in materials, the super-linear scaling with precision N for the numerical solution of its scattering integral term presents a formidable challenge and this has greatly restricted the applicability of the TDBE. Over the last few years, we have developed a novel solver for the solution of this very term, which greatly mitigates the scaling challenge. For instance, for a 3-leg phonon scattering description our solver shows a scaling with precision as N^2.5 in contrast to the N^4 scaling of conventional methods. This significant gain has already enabled us to successfully use the Boltzmann approach to decipher far-from-equilibrium ultrafast dynamics in upcoming quantum materials like CNTs and GeTe. In this project I propose to augment our solver with i) An Artificial Intelligence (AI) predictor of the morphology of the complex hypersurface involved in the scattering integrals of the TDBE and ii) an AI-assisted quadrature method. With these enhancements I envisage a further decrease in the scaling of the numerical cost to N^2 leading to a significant gain even at moderate precision, thereby allowing a detailed description of hydrodynamic phonon flow in graphene under a temperature gradient and even for complex geometries. The success of this project will not only result in a first-ever modelling of phonon Poiseuille flow in graphene and thin graphite but will also open exciting avenues for major technological advances through the realisation of ultra-high heat dissipators for modern electronics.