Hanchen Wang
@hcwww_
Followers
986
Following
199
Media
6
Statuses
163
chill @Stanford @Genentech ; with @Jure #Aviv
Joined May 2013
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#Eurovision2024
• 1443921 Tweets
América
• 489127 Tweets
New Jersey
• 221256 Tweets
#धरती_को_स्वर्ग_बनाना_है
• 181169 Tweets
#母の日
• 155711 Tweets
Nemo
• 120414 Tweets
ヴィクトリアマイル
• 62568 Tweets
#MothersDay
• 60490 Tweets
#UFCStLouis
• 47995 Tweets
Barco
• 30835 Tweets
Tatum
• 30010 Tweets
Pachuca
• 29430 Tweets
カーネーション
• 28767 Tweets
ストフェス
• 27108 Tweets
Igor
• 23763 Tweets
Saint MSG Insan
• 20112 Tweets
カグヤ様
• 19463 Tweets
Canes
• 17958 Tweets
Roger Corman
• 16886 Tweets
Garland
• 16316 Tweets
ナミュール
• 14064 Tweets
Buckley
• 11443 Tweets
Derrick Lewis
• 10937 Tweets
Today marks my day1 as a postdoc. In the following years, I’ll be at
@Genentech
and
@Stanford
, with
#AvivRegev
and
@Jure
, working on challenging and important problems. I’ve only applied this position upon graduation
@Cambridge_Uni
, really excited for this new journey!
12
21
362
I find that AI/ML papers published in Nature and Science may not always lead in engineering metrics or test scores. However, they are always the masterpieces in defining problems, solving them effectively, and presenting results impactfully and insightfully🧑🚀👩🚀
7
7
124
LK-99 highlights vast, under-discovered realms in chemistry/materials where AI might contribute. After years of dedication, our 14-page observations, insights and perspectives on
#AIforScience
is published in
@Nature
2
27
115
Honored to co-lead this project, and share our findings at
@NatMachIntell
!
An
#AI
model that can accurately diagnose
#COVID19
while preserving the privacy of patient data has been developed by a team from
@Cambridge_Eng
and
@FacultyMaths
:
#PublicHealth
1
14
41
0
1
11
Great work! As the main host for interns at Aviv's Lab next summer, I am now actively searching for more top talents to join us.📢 Intern
@Genentech
. We've already given out some offers, stellar students from prominent labs. Job description:
Heroic effort by our summer intern Anay Gupta,
@lauragunsalus
and
@lal_avantika
. Also the first preprint of our group, very proud!
0
8
67
1
7
47
We are looking for one more summer intern at Regev Lab,
@Genentech
🧬💊. We are interested in candidates who are experienced in spatial transcriptomics or have fascinating ideas on LLM for Genomics. Please apply through this link and drop me a message :)
Great work! As the main host for interns at Aviv's Lab next summer, I am now actively searching for more top talents to join us.📢 Intern
@Genentech
. We've already given out some offers, stellar students from prominent labs. Job description:
1
7
47
1
12
33
2nd Year of PhD ✅
almost all the projects of first year accepted 🆙🆙
0
0
22
Our "AI for Science" paper (, Nature 2023) is a good start. Some must-reads in my head include (Kaufmann, Nature 2023), (Degrave, Nature 2022), and (Bi, Nature 2023) for their field-expanding insights. For a specialized look at viral escape, Hie's work…
1
2
16
We will take care of the Point
#3
#AIforScience
🚀
Six predictions for AI in 2024:
- A hyped AI company will go bankrupt or get acquired for a ridiculously low price
- Open-source LLMs will reach the level of the best closed-source LLMs
- Big breakthroughs in AI for video, time-series, biology and chemistry
- We will talk much…
176
308
2K
0
0
13
We start by identifying the three stages of scientific discovery: observation, hypothesis, and experiments, we share successful cases. Next, we explore AI's role in facilitating these stages, concluding with its challenges and opportunities.
1
1
7
@vijaypande
human developed a weapon -> we should regulate the weapon; human used AI to develop a weapon -> we should regulate the AI 😓
1
1
6
My first project on cell imaging🔬🧬🧫, started during the 2nd year of PhD ('19-'22). There are some new exciting trends & techs worth checking👇🏻:
- OpenCell,
- JUMP,
- Roche/Genentech + Recursion,
Online now: Focalizing regions of biomarker-relevance facilitates biomarker prediction on histopathological images
0
0
2
2
0
5
SAM also enlightens "Foundation Models in Science"! Key recipe:
1. Promotable tasks that enable zero-shot generalization;
2. Models with flexible input, real-time amortizable;
3. Cost-effective data annotation via model-in-the-loop
#AIforScience
#FoundationModels
Today we're releasing the Segment Anything Model (SAM) — a step toward the first foundation model for image segmentation.
SAM is capable of one-click segmentation of any object from any photo or video + zero-shot transfer to other segmentation tasks ➡️
144
2K
7K
0
0
4
@Yubin_Xie
See you at NeurIPS! Have paper to present (hopefully) also have quite some experience on "failure mode of (my own developed) foundation models😅"
1
0
3
@simocristea
@mo_lotfollahi
I think zero-shot embeddings like 'Universal Cell Embeddings' are a great way to work with ever-growing amounts of data. They can save time in finding and refining the right data for fine-tuning on different datasets or tasks. It's similar to what word2vec and BERT did for NLP.
0
0
2
@DrJimFan
regarding AI + Science, it is worthing checking our recent piece published in Nature, we need more knowledge than LLM/Agent :)
0
0
2
@micahgoldblum
My fav part is
@kchonyc
's analogy drawing parallels between randomized controlled clinical trials and the deployment of deep learning amidst its interpretability challenges.
0
0
2
@ZimingLiu11
Thanks for sharing! Many submissions that focus on specific problems in Materials, Biology, or Chemistry could be categorized as 'Deep Learning for Science.’..
0
0
1
This review wouldn’t come out without
@TianfanFu
and the team’s commitments, especially
@YuanqiD
’s efforts in organizing
@AI_for_Science
workshops with
@NeurIPSConf
@icmlconf
etc. Kudos to
@marinkazitnik
who worked with us three for years long since the beginning.
1
0
2
Great work! I remembered the wonderful days we co-organised the “ML for Materials” at ICLR this year🚀
Today in
@Nature
, our team at
@GoogleDeepMind
is excited to share GNoME, a deep learning system that increases the number of stable crystal materials known to humanity by an order of magnitude.
From the data, we train a ML force-field with unprecedented capabilities at scale👇
8
49
363
0
0
2
@Xiaojie_Qiu
@JswLab
@hao_wu_7
@WangXiaoLab
@coletrapnell
@JShendure
@jhxing001
@bloodgenes
Congrats!!
1
0
1
@EricTopol
Thanks for sharing our work!
How can
#AI
transform science?
Let us count the ways
A brilliant review
@Nature
@marinkazitnik
@TianfanFu
@YuanqiD
and colleagues
@AI_for_Science
#ScienceTwitter
12
104
330
0
0
0
💡Reinforcement Learning ➕ Material Structure Prediction, from Levine’s Group
We can use deep nets not just to *predict* but to *optimize* -- train on data (e.g., experiments), and use the model to find designs that do better than the ones tested. We developed a new method for this model-based optimization problem that optimizes in a latent space. A 🧵:
4
43
270
0
0
1
@ml_angelopoulos
UCSF for sure, but I don’t have names in my mind. Also try compbio PIs in Cal? Nir Yosef, Yun Song, Jennifer Listgarten etc, they also know much more bio people in Bay Area.
1
0
1
@simocristea
@mo_lotfollahi
And I strongly agree that "FMs in single cell genomics are still in search for problems."👍
0
0
1
@JamesTneal
Outstanding work! It would be interesting to see how learning-based image encoders (e.g., neural networks) perform on the cell retrieval tasks.
0
0
1
@rsiva
@AI_for_Science
The 2nd refers to AoE (Anytime on Earth), where the 3rd on OpenReview is the GMT :)
0
0
1
@Yubin_Xie
@KempnerInst
According to Feifei’s talk today, probably no one has H100 yet , waiting to be shipped👀
1
0
1
@fjanoos
True, some are, for sure, artificial and skeptical or even misleading. Yet, keep open-minded with carefulness is no harm. Back to the Copernicus's time, 99.99% people believed that Earth is the central of universe. It is always the spirally developing science and engineering that…
0
0
1