Yuliang Leon Sun, MD, PhD retweeted
🎉 Our paper "Pre-trained Gaussian Processes for Bayesian Optimization" has been published in the Journal of Machine Learning Research! Check out a new way to connect human beliefs with model uncertainty predictions for decision making.
jmlr.org/papers/v25/23-0269.…
Yuliang Leon Sun, MD, PhD retweeted
Ever struggled with transferring knowledge in Bayesian optimization due to varying dimensions? Our TMLR paper solves this with MPHD! New Gaussian process Model Pre-training method works seamlessly across Heterogeneous Domains.
arxiv.org/abs/2309.16597
#MachineLearning
Thank you to all the co-authors for their contributions in this years long journey, showcasing dedication to our patients and advancing the scientific frontier.
We are thrilled to announce our latest publication out today @jclinicalinvest on iPSC modeling of childhood interstitial lung disease caused by ABCA3 mutations:
jci.org/articles/view/164274
Yuliang Leon Sun, MD, PhD retweeted
We are thrilled to announce our latest publication out today @jclinicalinvest on iPSC modeling of childhood interstitial lung disease caused by ABCA3 mutations:
jci.org/articles/view/164274
Yuliang Leon Sun, MD, PhD retweeted
Do we really need a gigantic amount of data to train AI models? We've been exploring this question from several facets and here's our first step to ensure robustness during data downsampling!
How to downsample data to safeguard against worst-case generalization performance issues (stemming from class imbalance and distributional shifts)? Great work with @bankes_william, G. Hughes and @ziwphd
Yuliang Leon Sun, MD, PhD retweeted
Using LLMs for a new domain-specific language?
We developed Grammar Prompting to teach them rules&structures, boosting their skills in semantic parsing, planning 🤖, and even molecule generation!
Excited to share this work #NeurIPS2023 poster #329 Tuesday at 5pm!
Replying to @ziwphd
@Ziwphd and I will present our Grammar Prompting work tomorrow at #NeurIPS2023
Happy to chat more on
- LLMs for structured language generation (e.g., programs, molecules and robotic plans)
- Structured/formal chain-of-thought for reasoning
Yuliang Leon Sun, MD, PhD retweeted
Introducing Gaussian Process Probes (GPP) for probing and measuring uncertainty about concepts represented by AI models. #NeurIPS poster #1504 this Tue at 5pm!
Paper: arxiv.org/abs/2305.18213
Joint work with @alex_y_ku, @jasonbaldridge, Tom Griffiths @cocosci_lab and @_beenkim
Yuliang Leon Sun, MD, PhD retweeted
This is a really natural framework to improve Bayesian optimization when you have access to related optimization tasks arxiv.org/abs/2109.08215
Joint work with @ziwphd, @GeorgeEDahl, Chansoo Lee, @zacharynado, @jmgilmer, @latentjasper, @ZoubinGhahrama1
Hyper Bayesian optimization (HyperBO) is a highly customizable interface that pre-trains a Gaussian process model and automatically defines model parameters, making Bayesian optimization easier to use while outperforming traditional methods. Learn more → goo.gle/3GnMYHG
Yuliang Leon Sun, MD, PhD retweeted
Where should Bayesian priors come from and how can we trust them for decision making? tldr is pre-training to align with expert beliefs!
arxiv.org/abs/2109.08215
Kudos to
@GeorgeEDahl,
@kswersk, Chansoo Lee, @zacharynado, @jmgilmer, @latentjasper, @ZoubinGhahrama1 for the collab!
Hyper Bayesian optimization (HyperBO) is a highly customizable interface that pre-trains a Gaussian process model and automatically defines model parameters, making Bayesian optimization easier to use while outperforming traditional methods. Learn more → goo.gle/3GnMYHG
Yuliang Leon Sun, MD, PhD retweeted
Check out github.com/google-research/h… to learn more about how to pre-train a Gaussian process!
This is a really natural framework to improve Bayesian optimization when you have access to related optimization tasks arxiv.org/abs/2109.08215
Joint work with @ziwphd, @GeorgeEDahl, Chansoo Lee, @zacharynado, @jmgilmer, @latentjasper, @ZoubinGhahrama1
Yuliang Leon Sun, MD, PhD retweeted
Our latest work just out @ajrcmb featuring a new ABCA3:GFP reporter for iPSC-derived alveolar type 2 cell purification. Special thanks to MD/PhD student @YLeonSun who led the project @CReM_Boston.
atsjournals.org/doi/10.1165/…
Yuliang Leon Sun, MD, PhD retweeted
I want to be able to look back on these years and say, my generation delivered climate solutions, racial equality, and an end to endless war. Help me deliver that new generation to Washington before it's too late. #DemDebate
Yuliang Leon Sun, MD, PhD retweeted
When people disappoint us, it’s not due to their actions. It’s because their actions fell short of our expectations.
You can’t control what people do, but you can choose not to let their actions drive your emotions.
Be clearer about your expectations—or set more realistic ones.
Yuliang Leon Sun, MD, PhD retweeted
Just because an idea is popular doesn't mean it's effective.
In open offices, people actually interact less.
Learning styles don't predict learning.
10,000 hours of practice doesn't always make perfect.
Fadbusting with @Freakonomics, a #WorkLife bonus: smarturl.it/WorkLifeseason2
Yuliang Leon Sun, MD, PhD retweeted
Gaining followers doesn't make you a leader. It just means you've captured attention.
Leadership is the art of uniting individuals to achieve a collective goal.
Great leadership is the art of elevating a group to achieve a worthy goal.
#FridayFeeling
Yuliang Leon Sun, MD, PhD retweeted
Pete Buttigieg says an agenda for black Americans needs to include "five things that all of us care about: homeownership, entrepreneurship, education, health, and justice." n.pr/2VqBBoK
Yuliang Leon Sun, MD, PhD retweeted
Last day of the iPS Course at the CReM! #opensourcebiology #stemcells #regenerativemedicine #wearetheCReM
Yuliang Leon Sun, MD, PhD retweeted
We should respect every family of every race, creed, or religion who loses a loved one due to violent crime. But the President of the United States uses his powerful platform to weaponize “Angel Families" and demonize undocumented immigrants. Horrifying.
Yuliang Leon Sun, MD, PhD retweeted
Grad students, here is an outline of how to give your first thesis committee meeting:
1-What is your question?
2-Why significant?
3-Overview of what is known and gaps in knowledge
4-Approach
5-Progress or challenges
6-Future plans
Remember, it's a conversation not an exam!



















