Jinlin Lai

Jinlin Lai

MS/PhD student at University of Massachusetts Amherst

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Dessert - the alien decipher

4 minute read

Published:

(Cryptologist, calm down when I seem overconfident about what I don’t know)

Appetizer - the cruel zoo

6 minute read

Published:

(Readers under the age of 13, please read with your guardian)

Preface

3 minute read

Published:

I finally start to translate some of my high school works here. It is not because I have nothing else important to do, but to commemorate the positive parts of my messy adolescent. I wish I will always have a teenager’s mind, and act as a serious adult, till my end. While translating, I realized a lot I have not noticed before. So the writings and thoughts are a lot more mature than their original forms, which may make it seem overstated that a high schooler can figure all those out. However, the core ideas were indeed formed around 2014, when I was a fairly active thinker of everything I learned.

portfolio

publications

Shallow VAEs with RealNVP Prior Can Perform as Well as Deep Hierarchical VAEs

Published in The 27th International Conference on Neural Information Processing (ICONIP2020), 2020

This project inspired my interest in machine learning research. Although it was rejected twice due to lack of contribution, it is a very complete work. I am proud of contributing to it.

Recommended citation: Xu, H., Chen, W., Lai, J., Li, Z., Zhao, Y., & Pei, D. (2020). Shallow VAEs with RealNVP Prior can Perform as Well as Deep Hierarchical VAEs. ICONIP. https://link.springer.com/content/pdf/10.1007%2F978-3-030-63823-8_74.pdf

Optimal Mixture Weights for Off-Policy Evaluation with Multiple Behavior Policies

Published in Offline Reinforcement Learning Workshop at Neural Information Processing Systems, 2020

This is my first paper as first author. It is the theory part of my undergraduate dissertation. The paper was rejected by NeurIPS 2020 due to strict assumptions and poor writings. To be honest, I was surprised that the reviewers are not picky about the experiments. They actually indicate that our methods only apply to very short horizon.

Recommended citation: Lai, J., Zou, L., & Song, J. (2020). Optimal Mixture Weights for Off-Policy Evaluation with Multiple Behavior Policies. arXiv preprint arXiv:2011.14359. https://arxiv.org/pdf/2011.14359

Variational Marginal Particle Filters

Published in The 25th International Conference on Artificial Intelligence and Statistics (AISTATS) 2022, 2022

This is my first publication as first author. I enjoy working on it, especially when developing the theories. After the publication, I have made up my mind to contribute more on the theory aspects of machine learning.

Recommended citation: Lai, J., Domke, J., & Sheldon, D. (2022). Variational Marginal Particle Filters. _AISTATS_. https://arxiv.org/pdf/2109.15134.pdf

Automatically Marginalized MCMC in Probabilistic Programming

Published in The 40th International Conference on Machine Learning, 2023

This is my first publication in the field of probabilistic programming. Hope you find it interesting! I personally think that every PPL in the world that has an HMC pipeline should implement this idea.

Recommended citation: Lai, J., Burroni, J., Guan, H., & Sheldon, D. (2023). Automatically Marginalized MCMC in Probabilistic Programming. _ICML_. https://arxiv.org/pdf/2302.00564.pdf

talks

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.