Charles Yang

About

Hi, I'm Charles Yang! I don't know how you ended up on my little island floating out in the vast internet, but welcome! I'm a Christian (Christ-follower). I know beliefs like this seem antiquated in our time, but I have found that Christianity provides a wellspring of meaning and purpose for my life. But it is more than just a convenient philosophy, its claims are actually grounded in real, historical events (1 Corinthians 15:14) that have radical implications for how we live our lives. If you have questions or want to learn more, I'd love to talk to you about it!

In my free time, I’m also a 5th year masters student at UC Berkeley in the Electrical Engineering and Computer Science (EECS) department, focusing on AI and dynamical systems. More broadly, I'm interested in applying machine learning and deep learning to various scientific applications. I'm currently interning at Lawrence Berkeley Lab and conducting research under Grace Gu in the Mechanical Engineering Department at Berkeley. I'm also working in Michael Mahoney's research group applying dynamical control theory to recurrent neural networks (RNN). Recently, I started a newsletter focusing on applications of machine learning to various scientific fields called ML4Sci.

Trust in the Lord with all your heart, and do not lean on your own understanding. In all your ways acknowledge Him, and He will make straight your paths. Be not wise in your own eyes; fear the Lord, and turn away from evil.
Proverbs 3:5-7

News

Selected Publications[Google Scholar]

Figure Abstract
Prediction of composite microstructure stress-strain curves using convolutional neural networks
Materials & Design, 2020
Charles Yang, Youngsoo Kim, Seunghwa Ryu, and Grace X. Gu
Figure Abstract
Using CNNs to predict composite properties beyond the elastic limit
Material Research Society Communications (MRS Comm.), 2019
Charles Yang, Youngsoo Kim, Seunghwa Ryu, and Grace X. Gu

Power grid voltages after PV installation
Sequential Optimal Placement of Distributed Photovoltaics using Downstream Power Index  
North American Power Symposium 2017 (NAPS 2017)
Mir Hadi Athari, Charles Yang, Zhifang Wang

Journal cover image
Facile synthesis of ZnO@ZIF core–shell nanofibers: crystal growth and gas adsorption  
Crystal Engineering Communication (Cryst. Eng. Comm.), 2017
Xiang He, Charles Yang, Dawei Wang, Stanley E. Gilliland III, Da-Ren Chena, and Wei-Ning Wang

Blog[Medium][Substack]

Thoughts on COVID-19, Scientific Gatekeeping, and Substack Newsletters

COVID-19 has laid bare the inadequacies and fault lines in the way we communicate science to the general public. In this essay, I explore how the machine learning community has suffered from similar problems and potential solutions to improve the way we communicate and disseminate scientific research. For a follow-up to this essay that covers recent developments in the open science community, see ML4Sci #15


Defining the new AI-powered SaaS: Science as a Service
Distributed by Artificial Intelligence curators at Medium!
Published in The Startup!

As I’ve been discussing throughout my newsletter ML4Sci, the intersection of AI and science has particular nuances that make its adoption and development subtly different from the development of AI as a whole - the same is true of its business models. In this essay, I’ll be stepping out of the academic world of ML4Sci and instead examine how AI is shaping the business realm for science-based startups and companies. I will argue that "software is eating up" science and as a result, we will see new cloud-based SaaS models as ML4Sci becomes industrialized and commercialized.


Interviewing the 1.5B GPT-2 model by OpenAI
Distributed by Data Science and Machine Learning curators at Medium!
Published in Towards Data Science!

Using the 1558MB version of OpenAI’s GPT-2 model and Max Woolf’s gpt2-simple package on github powered by Google colaboratory, I gave GPT-2 a series of data-science focused prompts and analyzed their responses. I also analyze the generated text and discuss implications of such powerful language models.


ML4Sci: #1: Discovering new materials from abstracts; Designing diffractive metagratings with GAN's; How ML can help fight climate change; +Industry Highlight

Check out the inaugral issue of ML4Sci, my newsletter on applications of machine learning in various scientific fields.

Deep Learning in Science
Distributed by Machine Learning, Science, and Data Science curators at Medium!
Published in Towards Data Science!

Deep Learning has traditionally been motivated by the concerns of tech companies such as Facebook and Google. However, it also has many applications in scientific fields, ranging from molecular chemistry to astrophysics. While there are subtle differences in implementation and different data structures, Deep Learning has the potential to fundamentally alter the way we do science. In this post, I introduce Deep Learning in the context of scientific and engineering applications, consider the opportunities for using Deep Learning, and discuss different patterns of Deep Learning for scientific problems.


Achieving Mathematical Maturity
Distributed by Math curators at Medium!
Published in Towards Data Science!

"Mathematical maturity" is an often-used catch phrase in higher education that it is rarely defined, and even more rarely is one told how to achieve it. In this post, I share some insights I've gained on how one should approach learning math, from my experience in both math classes and various technical classes that use math.


Metal Organic Frameworks - A Brief Introduction
Distributed by Science curators at Medium!

Metal Organic Frameworks (MOF) are one of the most exciting classes of materials discovered in the 21st century. MOF's are versatile not only in potential applications, ranging from catalysis, to gas storage, to thereapeutic drug delivery, but also in terms of functionality and synthesis. Composed of metallic nodes and organic linkers, a wide variety of combinations of transition metals and simple organic compounds can be used to synthesize MOF's in a variety of environments. In this post, I discuss various synthesis methods, properties of MOF's, and some example applications of MOF's. All included articles are open-source!