RESEARCH INTERESTS

Research background: My research journey started at National Taiwan University (NTU) when I was a master’s student working with Prof. Yeong-Bin Yang on vibration analysis of a symmetric two-member truss. This truss system can be described by a simple governing equation, but it can display chaos phenomena when geometric nonlinearity is considered. I was fascinated by how a deterministic system can generate complexity from simplicity, known as the butterfly effect. This early research experience motivated me to pursue a Ph.D. degree in Structures and Materials at MIT. While I studied the finite element method (FEM) and structural dynamics at NTU, my research at MIT focused on atomistic modeling and quantum chemical calculations. I worked with Prof. Markus Buehler to investigate the structural, mechanical, and optical properties of eumelanin (natural pigment found in most organisms) and polydopamine (mussel-inspired material for multifunctional coatings) using multiscale modeling. Multiscale modeling is a technique that combines various simulation methods to capture the behavior of materials at different scales (both in length and time), providing links connecting the atomic-scale and macro-scale phenomena. By analyzing the collective behavior of the atoms in a material system, its structure-property relationship could be revealed. At MIT, I developed research expertise in atomistic modeling, including molecular dynamics (MD) simulations and density functional theory (DFT). The research findings in my thesis were published in high-impact journals, including ACS Nano, Nature Communications, and Chemical Science, and highlighted on MIT News.

Passion for artificial intelligence: Inspired by the success of Google DeepMind’s AlphaGo program in 2016, I saw the potential of AI and ML to advance scientific computing. During my postdoctoral training at MIT, I published two papers on data-driven composites design. The first one was in Extreme Mechanics Letters and the second one was in Materials Horizons and received the Materials Horizons Outstanding Paper Award. Both papers were published in 2018 and have received more than 250 citations as of today (Dec 2022). During my postdoctoral training at UC Berkeley, I published four ML-related papers. In 2019, we showed that high-performing designs of graphene nanocomposites can be generated using a data-driven approach (published in Advanced Theory and Simulations as the cover) and wrote a prospective paper on ML for composite materials (published in MRS Communications) and received the MRS Communications Lecture Award. In 2020, we developed a general-purpose inverse design approach using generative inverse design networks (GIDNs). This work (published in Advanced Science) has inspired many follow-up studies around the world in different fields and received more than 120 citations as of today. In 2021, we developed ElastNet, a de novo elastography method combining the theory of elasticity with a deep-learning approach. This work (published in PNAS) shows that representing a target physical quantity field as a continuum and differentiable function by a neural network offers numerous benefits over conventional discrete grid-based representations, such as FEM.

 

I deeply appreciate the guidance from my former advisors and mentors at NTU (Yeong-Bin Yang), MIT (Markus Buehler), and UC Berkeley (Daryl Chrzan and Mark Asta)I also appreciate the financial support that I have received for my study and research. During my Ph.D. study at MIT, I received support via a fellowship from the Taiwanese Government and funding from CRP Henri Tudor in the framework of the BioNanotechnology project. During my postdoctoral training at MIT, I received support from the Office of Naval Research (ONR) and the Multidisciplinary University Research Initiative (MURI). During my postdoctoral training at UC Berkeley, I was fully funded by the Department of Mechanical Engineering.


Abstract: Elastography is an imaging technique to reconstruct elasticity distributions of heterogeneous objects. Since cancerous tissues are stiffer than healthy ones, for decades, elastography has been applied to medical imaging for noninvasive cancer diagnosis. Although the conventional strain-based elastography has been deployed on ultrasound diagnostic-imaging devices, the results are prone to inaccuracies. Model-based elastography, which reconstructs elasticity distributions by solving an inverse problem in elasticity, may provide more accurate results but is often unreliable in practice due to the ill-posed nature of the inverse problem. We introduce ElastNet, a de novo elastography method combining the theory of elasticity with a deep-learning approach. With prior knowledge from the laws of physics, ElastNet can escape the performance ceiling imposed by labeled data. ElastNet uses backpropagation to learn the hidden elasticity of objects, resulting in rapid and accurate predictions. We show that ElastNet is robust when dealing with noisy or missing measurements. Moreover, it can learn probable elasticity distributions for areas even without measurements and generate elasticity images of arbitrary resolution. When both strain and elasticity distributions are given, the hidden physics in elasticity—the conditions for equilibrium—can be learned by ElastNet.

Full paper: C-T Chen and GX Gu, Learning hidden elasticity with deep neural networks, Proceedings of the National Academy of Sciences, 2021, DOI: 10.1073/pnas.2102721118



Updated: Nov 9, 2020


Abstract: Atoms are the building blocks of matter that make up the world. To create new materials to meet some of civilization’s greatest needs, it is crucial to develop a technology to design materials on the atomic and molecular scales. However, there is currently no computational approach capable of designing materials atom-by-atom. In this study, we consider the possibility of direct manipulation of individual atoms to design materials at the nanoscale using a proposed method coined “Nano-Topology Optimization”. Here, we apply the proposed method to design nanostructured materials to maximize elastic properties. Results show that the performance of our optimized designs not only surpasses that of the gyroid and other triply periodic minimal surface structures, but also exceeds the theoretical maximum (Hashin–Shtrikman upper bound). The significance of the proposed method lies in a platform that allows computers to design novel materials atom-by-atom without the need of a predetermined design.

Full paper: C-T Chen, DC Chrzan, GX Gu, Nano-topology optimization for materials design with atom-by-atom control, Nature Communications, 2020, DOI: 0.1038/s41467-020-17570-1


Updated: Nov 9, 2020


Abstract: In recent years, machine learning (ML) techniques are seen to be promising tools to discover and design novel materials. However, the lack of robust inverse design approaches to identify promising candidate materials without exploring the entire design space causes a fundamental bottleneck. A general‐purpose inverse design approach is presented using generative inverse design networks. This ML‐based inverse design approach uses backpropagation to calculate the analytical gradients of an objective function with respect to design variables. This inverse design approach is capable of overcoming local minima traps by using backpropagation to provide rapid calculations of gradient information and running millions of optimizations with different initial values. Furthermore, an active learning strategy is adopted in the inverse design approach to improve the performance of candidate materials and reduce the amount of training data needed to do so. Compared to passive learning, the active learning strategy is capable of generating better designs and reducing the amount of training data by at least an order‐of‐magnitude in the case study on composite materials. The inverse design approach is compared with conventional gradient‐based topology optimization and gradient‐free genetic algorithms and the pros and cons of each method are discussed when applied to materials discovery and design problems.

Full paper: CT Chen and GX Gu, Generative deep neural networks for inverse materials design using backpropagation and active learning, Advanced Science, 2020, DOI: 10.1002/advs.201902607