Tianyu Liu

BS @ #UIUC, PhD student @ #Yale

[Tianyu Liu @meSuper8]

🎵Glad to share our work (accepted by #NeurIPS2023) on using GNN for gene embedding analysis with @WangYuge4585 , Rex Ying and @HongyuZhao2. To efficiently integrate cross-modal relationships in multi-omics data, we propose a novel method to learn the representation of genes.

[Yale Kavli Institute for Neuroscience @KavliAtYale]

Spotlight on Yong-Hui Jiang, MD, PhD, @YaleGenetics & @YaleNeuro. His team is developing a genome-editing🧬platform that could implement targeted therapies for neurodevelopmental disorders directly in the brain🧠

Read the project & Jiang's journey👇
https://t.co/U2yLrcNMjL

[Siebel School of Computing and Data Science @siebelschool]

A huge congrats to Tandy Warnow on her well-deserved election to the American Academy of Arts and Sciences!

As a prof of CS, assoc. dir. of the Siebel School and the Grainger Distinguished Chair in Engineering, we're thrilled to see her talent recognized!
https://t.co/cbviotlw4F

[Tianyu Liu @meSuper8]

Many thanks to my advisor Prof. Hongyu Zhao @HongyuZhao2 and other collaborators Dr. Xiangyu Zhang, Dr. Jiecong Lin @JasonLinjc , Prof. Pinello @lucapinello and Prof. Rex Ying for their contributions!

[Tianyu Liu @meSuper8]

Happy to see our method discussing the genomic language model pretraining with human variants (UKBioBERT) as well as model fusion (UKBioFormer) has been accepted by NPJ Artificial Intelligence: https://t.co/bGm0uy78DB

[Tianyu Liu @meSuper8]

We find that this approach can help us better classify genes, and predict cell-type-specific and individual-level gene expression levels. Model weights are returned to UKBioBank for accessing!

[wengong jin @WengongJin]

Thanks to NVIDIA's generous support, the best two papers at the ICML GenBio workshop will receive one DGX Spark each! We look forward to seeing your submissions.

[Faisal Mahmood @AI4Pathology]

📣 We are excited and thrilled to announce APOLLO, a healthcare system-scale multimodal temporal foundation model for virtual patient representations.

Trained on 25 billion clinical events from 7.2 million patients across 33 years and 28 modalities, APOLLO learns a unified atlas of medicine. Turning labs, notes, pathology images, medications, and diagnoses into coherent, computable longitudinal trajectories. APOLLO is disease-agnostic by design, a single model that learns the shared structure underlying human health and disease across every specialty, modality, and stage of care.

The possibilities are enormous: earlier risk prediction, treatment response modeling, clinical trial matching, biomarker discovery, and a new generation of agentic systems built on rich patient representations.

Read the pre-print: https://t.co/IUaWyodcrS
Read our blog post about the work: https://t.co/QOeUHYsrYN

👏 🎉Huge congratulations to Andrew Zhang , @TongDing99, Sophia J. Wagner, and the rest of the team.

[Violet Peng @VioletNPeng]

Congrats @ysu_nlp for this exciting launch! Cannot wait to see what will come out from @NeoCognition!

[Tianyu Liu @meSuper8]

Michelle is fantastic!

[Michelle M. Li (李敏蕊) @DrMichelleMLi]

I am super excited to share that I am joining @CarnegieMellon as a tenure-track Assistant Professor in Biomedical Engineering @cmu_bme 🎉

[Anthropic @AnthropicAI]

Research we co-authored on subliminal learning—how LLMs can pass on traits like preferences or misalignment through hidden signals in data—was published today in @Nature.

Read the paper: https://t.co/b1BYwcW9dH

[wengong jin @WengongJin]

Excited to announce the third GenBio workshop at ICML 2026! You are welcome to submit your work! https://t.co/Fc3v5yXHgH

[Stanford NLP Group @stanfordnlp]

For this week's NLP seminar, we are excited to host @hhsun1 from The Ohio State University!

Date and Time: Thursday, April 16, 11:00AM — 12:00 PM Pacific Time.
Zoom Link: https://t.co/jmz2wb8Xyn

Title:  Smarter and Safer Every Day? Continual Learning and Safety in Computer-Use Agents

Abstract: Computer-use agents are increasingly deployed in open-ended digital environments — interacting with the web and executing multi-step tasks across applications. Yet real-world deployment exposes a fundamental tension: agents must continually adapt to unseen scenarios and distribution shifts, while their safety and security grow more critical as their capabilities increase. In this talk, I will discuss our recent efforts toward continual learning in computer-use agents for environment adaptation, systematic evaluation of adversarial risks, and proactive discovery of unintended behaviors in the absence of attacks. Together, these efforts suggest that capability and safety need not be in tension, and point toward agents that are both capable and safe in deployment.

Hope to see you all there!

[Tianyu Liu @meSuper8]

Check our great work! It is my pleasure to work with Yuanqi and other team members, and I am happy to see the power of AI agents in creating new objectives and producing new drugs, cis-regulatory elements, and materials!

[Yuanqi Du @YuanqiD]

❓How can we build AI agents that do what scientists actually do? Is scientific discovery merely a search problem?

🚀 Meet SAGA: Scientific Autonomous Goal-evolving Agents. Five discovery tasks across chemistry, biology & materials science, with wet-lab validation.

[Yuanqi Du @YuanqiD]

I am excited to share what I have been learning about LLMs and its possibility to automate scientific discovery this Friday! We have pretty impressive results along the way! Come that interests you!

[Tianyu Liu @meSuper8]

@BoWang87 two tokens=二词元

[Tianyu Liu @meSuper8]

This paper has been published after a lengthy peer-review process, and I really appreciate the support from my advisor Prof. Zhao @HongyuZhao2 as well as other co-authors Mr. Kexing Li, Dr. Yuge Wang, and Dr. Eddie Li!

[Tianyu Liu @meSuper8]

In scEval, we attempted to develop some of the earliest task-driven benchmarking schemes to evaluate the performance of various single-cell foundation models. https://t.co/NOWHBQ4n0O

[Tianyu Liu @meSuper8]

We have considered 10 different models from different time peroids, it is nice to see the pros and cons of these methods, and we hope the conclusion can help us develop better models for virtual cell study!

[Forbes @Forbes]

Pharmaceutical giant Roche announced it is launching a new large-scale “AI factory” with thousands of the latest Nvidia chips to accelerate the development of new drugs and diagnostics.

Roche said the company has ramped up its AI capacity by purchasing 2,176 Nvidia Blackwell GPUs, which will be deployed across the U.S. and Europe.

That brings its total GPUs to more than 3,500, which it claims is the greatest number owned by any pharmaceutical company.

Roche, which made the announcement in conjunction with Nvidia’s GTC conference in San Jose, hopes to use that technological firepower to discover, develop, manufacture and commercialize therapies faster.

Learn more: https://t.co/IHHZFyFcVT (📸: Richard Morgenstein 2021)

[Tianyu Liu @meSuper8]

@ZhiYu_ACGT @HongyuZhao2 @YuvalKluger Thank you Zhi

[Tianyu Liu @meSuper8]

Finally, be Dr. Liu. Many thanks to my advisor, Prof. Zhao @HongyuZhao2, for his unwavering support; he is my role model, for sure! Also, thanks to my committee members Prof. Kluger @YuvalKluger and Prof. Ying, as well as my collaborators, friends and family!