Learning Unlimited Cloud

Splash Biography

MICHAEL KIELSTRA, Student mathematician; library enthusiast

Major: Not available.

College/Employer: Not available.

Year of Graduation: 2022

Picture of Michael Kielstra

Brief Biographical Sketch:

I'm a junior at Harvard studying mathematics and working on reinforcement learning under Dr. Susan Murphy. (Google her. Google me too -- I'm much less impressive but it makes me happy to go on the analytics page and see that people clicked on my website.) Ask me about algebra, calculus, Harvard, college admissions, computer science, the Boy Scouts, science fiction, and navigating by the stars.

Past Classes

  (Clicking a class title will bring you to the course's section of the corresponding course catalog)

H471: The Law as Literature: A Close Reading of Fraudulent Legal Arguments in Rainstorm Fall 2020 (Dec. 05 - 06, 2020)
"The law" is surprisingly hard to pin down as a concept. It is a mass of Acts of Congress, judicial decisions, and annotations, written in a language which often seems very different from any other. Understanding the meaning of the law often seems to be the domain of those who have spent years studying it. In this class, we will take a different approach, reading legal decisions from a literary perspective and debating, not what they say about what is and isn't legal, but what they say about the nature of the people who were involved in the cases. We will, in short, apply literary criticism to the law, in the hopes of demonstrating that English class and Civics class are not as separate as they appear to be. In the process, we will attempt to answer the single most critical question of government: what is the purpose of having any law at all? Our focus will be the 2012 Canadian case of Meads v. Meads. In 2012, Dennis Meads's ex-wife sued him for child support. He claimed that he could not be convicted by the court: since it displayed Canada's crest, complete with motto "From sea to sea", it was in fact an Admiralty court and had no jurisdiction on land. He seemed completely certain that this was a good idea, and further investigation revealed that he had been scammed by someone posing as a law expert and selling him unworkable arguments. We'll discuss the ways in which this scam is more intricate than it appears, and what its existence, and the judge's response to it, says about the law in Canada and eleswhere. You don't need to know any law to take this class.

M24: No No No Yes: An Introduction to Reinforcement Learning in Rainstorm Spring 2020 (May. 30 - 31, 2020)
Computers do exactly what we tell them, but sometimes we don't know what to say. Some environments are just too complex or mysterious for a programmer to write down an explicit algorithm. Reinforcement learning is how computers learn to make their own decisions in a human-like way: trying, failing, practicing, and learning. It's used by medical professionals, by data scientists, and by Netflix's recommendation algorithms. In this fast-paced class, split between theory and practice, we will first study the design of, and then build, our own reinforcement learning algorithms. And get your bandanas ready, because we might just have a visit from the Bandit...